Start with the free vocal chain
Every guide here assumes a vocal that is already set up sensibly. The free StudioRescue chain does that part for you with REAPER's stock plugins: high-pass, corrective EQ, compression, de-essing and a reverb send, pre-wired. Stock plugins only, nothing to buy.
Get the free vocal chainPlugins missing after opening a REAPER project on another computer? Here is the fix
When a REAPER project opens on a second computer with plugins missing, the article explains that the project file records which plugins sit where and with what settings but does not contain the plugins, so anything not installed on the new machine cannot load, while the audio is unaffected. It quotes the official user guide that the project still loads and the FX stays listed in the chain but is not available, and warns against deleting those entries: save a copy first, and treat keeping the plugin's settings with the entry as StudioRescue's understanding to test rather than a documented fact. It lists what is missing from each FX chain and from the FX page of the Project Media/FX Bay, then installs the same plugin in the same format (VST, VST3 or CLAP prefix), adds its folder and re-scans, and reopens. A plugin that crashes the project is isolated with Open with FX offline (recovery mode). When a plugin cannot be installed, the options are freezing or applying FX on the original computer, or replacing it by ear. For next time: Save Project As with copy all media, freeze tracks before moving, lean on stock plugins, and send stems to other people. It is clear about the limits: no fix for OS support, licensing, identical replacement sound, or offline media.
Audio Repair and RestorationOnly have a low-bitrate MP3? How to improve it in REAPER (and what cannot be fixed)
For a recording that only survives as a low-bitrate MP3, the article is plain that the detail the encoder discarded cannot be restored, and that converting to WAV or raising the sample rate changes nothing about the sound. It explains the two kinds of damage, the lowpass ceiling that LAME's documentation says depends on the target bitrate and the artifacts left in what remains, then puts finding a better copy first: the recording device, the other person's local file, the platform, or an older WAV export. Three checks follow: read the bitrate, find the ceiling with REAPER's per-item spectrogram (Spectral edits, Show spectrogram), and listen to sibilants and pauses for shimmer. To avoid a second round of loss it works in WAV and encodes to MP3 only once for delivery. The stock ReaEQ order is a high pass around 80 to 120 Hz, a high shelf cut of about 2 to 6 dB or a low pass just under the ceiling to tame the swirl, a small presence boost only where content exists, light noise reduction at most, and level last. It describes iZotope's Spectral Recovery as re-synthesizing rather than recovering high frequencies, untested by StudioRescue, and lists what none of this will fix.
Home Vocal Recording and MixingWhy your vocal sounds good soloed but disappears under the beat (and how to fix it in REAPER)
When a vocal sounds clear soloed but disappears once the beat plays, the article identifies the usual cause as masking rather than level: the beat covers a frequency range the voice needs, so raising the fader makes the vocal louder without making it clearer. It gives three tests before any EQ: raise the vocal about 3 dB to see whether it is simply a balance problem, mute parts to find the one covering the voice, and check in mono with the master track's Mono/Stereo button, switched back afterwards. Using the official REAPER user guide's frequency overview, which suggests lowering 2 to 4 kHz a little on the mix while boosting a little on the vocal, it sets a starting method: a wide, shallow ReaEQ cut of about 1.5 to 3 dB on the beat, swept between 2 and 5 kHz with the vocal playing, a small matching lift on the vocal only if needed, and optional light sidechain ducking set up the way the guide's section 17.5 describes, with ReaComp's Detector input on Auxiliary Input L + R and a ratio between 2:1 and 4:1 kept to a dB or two of reduction. It is plain about the limits of working on a single stereo beat file and about what EQ cannot fix: a dull recording, an arrangement that fights the voice, unclear diction, or a beat that is simply too loud.
REAPER Audio WorkflowsHow to batch rename items, regions and markers in REAPER
The article separates the three names a beginner confuses: the take name shown on the item, the source file name on disk, and region and marker names on the timeline. For one name on many items it gives the stock route from the official user guide: select them, press F2 for Media Item Properties and type in the take name box, and it explains why matching names matter through REAPER's Propagate item to similarly named items commands. For dozens of regions or markers it gives the CSV round trip: export from the Region/Marker Manager, edit only the names, and bring the file back with Import replace rather than merge, which the changelog says leaves existing markers in place, with an undo point since REAPER 7.66 and a warning that take markers export but do not import. It is plain that neither the user guide nor the changelog documents a one-step way to give items numbered names, points to free community scripts for that with the guide's Load ReaScript route, covers renaming files on disk and track names, and advises naming tracks before recording so recorded files are named for you.
Podcast and Spoken-Word AudioHow to record a Zoom or phone interview into REAPER (when the guest cannot record their end)
When a remote guest cannot record their own end, the article keeps the goal the same as a double-ender: the host's microphone recorded cleanly in REAPER and the guest's call audio on a separate track, never one mixed file. It starts with Zoom's documented Record separate participant audio files option (up to 80 participants, phone dial-ins merged into one file), recorded alongside the host's own mic in REAPER with a clap at the start and end. For capturing the call inside REAPER it gives REAPER's WASAPI shared loopback mode on Windows, with the user guide's own beginner and hearing warnings, or an interface loopback feature, and on a Mac the free BlackHole driver with a Multi-Output Device and an Aggregate Device clocked to the interface. It covers echo prevention, phone-band quality, asking consent before recording, and is plain that every route still delivers call quality for the guest.
Practical MixingHow loud should you monitor when mixing (and why mixing loud fools you)
The article's position is stated in the first paragraph and it is a refusal rather than a number: no published standard tells a music mixer how loud to listen, and the official REAPER user guide contains no occurrence of dB SPL, sound pressure level, listening level or monitoring level across its 464 pages, so there is no setting in the program to find and no official figure to hit. What the level does change is every tone decision made at it, and the mechanism is given from the standard that governs it. ISO 226, Normal equal-loudness-level contours, is in its third edition as of March 2023 and specifies, across 20 Hz to 12,500 Hz, which sound pressure levels at each frequency are heard as equally loud; the structural point the article draws out is that the standard specifies a separate contour for each loudness level rather than one curve, because the shape of the ear's frequency response is not the same at every volume. Stanford's CCRMA course material on the equal-loudness curves supplies the direction: the curves flatten out at higher loudness levels, with the spread across the frequency range narrowing from roughly 70 dB at the lowest level to about 37 dB at a medium level and 28 dB at the highest. The practical consequences are drawn as three symptoms, a mix judged loud that under-provides bass and top and comes back thin and dull when played quietly, a mix judged very quietly that overcompensates and comes back boomy and harsh, and a level moved mid-session so that the second half of the decisions answer a different question from the first. A caveat follows immediately and is not softened: ISO 226 was measured with pure tones, a single frontal source, a free-field condition and otologically normal listeners aged 18 to 25, so the contours explain the direction of the effect and are not a correction curve to dial into an EQ. The 85 dB figure that circulates as a professional target is then repositioned as a ceiling. NIOSH's recommended exposure limit is 85 dBA averaged over an eight-hour shift, noise above 85 dBA is described as hazardous, and the exchange rate halves the allowable duration for each 3 dBA increase; the WHO and ITU safe-listening standard for personal listening devices is lower still, at 80 dB for 40 hours a week for adults and 75 dB for children. Set against a mixing week of roughly 40 hours of deliberate listening, the article argues that mix at 85 is the line rather than the benchmark, and offers a moderate conversational range as a starting point with a test that needs no meter, whether a conversation can be held over the music without raising the voice. The method is four steps and costs nothing: play a finished record known well and set the volume to where it sounds right rather than exciting, set it with the hardware control rather than REAPER's master fader, mark the position so it can be returned to next week, and make every tone decision there. The REAPER-specific section explains why the master fader is the wrong control, because the render takes the output of the master track and so passes through that fader on the way to the file, and gives the right place instead: the hardware output monitoring chain opened with View, Monitoring FX or Shift and the Master track FX button, which the official user guide states does not show up in renders, does not affect the metronome and is not stored in the project file, and which the guide itself suggests for running analysis FX or for compensating for the acoustic characteristics of a particular room. The catch is named rather than buried, that because the chain is not stored in the project it is global, so a trim set months ago still applies to every project opened. Measurement is covered with NIOSH's own tested limits on phone apps, 10 of 130 iOS apps meeting their selection criteria, three A-weighted apps within plus or minus 2 dBA of a reference meter which NIOSH noted matches the accuracy specified for Type 2 instruments, plus or minus 1 dB with external calibrated microphones, and only 4 of 62 Android apps partially meeting the criteria, with the honest boundary that this is enough to tell a reasonable level from a damaging one and not enough to call a room calibrated. Headphones are separated out as a different measurement problem, because the contours assume a frontal source in a free field, which leaves the repeatable part of the advice as the usable part. A table assigns decisions to levels, balance judged at the marked level and then quieter, low end at the marked level with a brief louder check, harshness and sibilance at the marked level and again after a break, and final loudness not by ear at all. The limits section states that a consistent monitoring level will not fix room modes, will not make a mix translate on its own, will not fix a balance that is wrong at every level, will not un-tire a fatigued ear inside a session, and will not produce a number to hit. Every figure is traced to the ISO catalogue entry, Stanford CCRMA, NIOSH, the WHO-ITU safe listening standard or the official REAPER user guide v7.81, checked on the day of publication against the current release v7.82 of 4 October 2026. No listening tests, room measurements or SPL readings are claimed, all levels are given as starting points or published limits rather than settings that will be correct in any particular room, and no link is an affiliate link.
REAPER Audio WorkflowsHow to use REAPER's undo history to get back an edit you lost
The article treats a lost edit as a navigation problem rather than a disaster, and its first move is to replace repeated presses of Ctrl+Z with a window most beginners never open. REAPER's Undo History window lists every undo point in the project as a readable sequence, and double-clicking an event loads that state directly, which is the difference between moving to a specific named point and stepping back blind one action at a time. Three routes in are given and all are equivalent: Edit, Undo History from the main menu, the Ctrl+Alt+Z toggle on Windows and Linux, and a right-click on either the undo or the redo arrow in the main toolbar. The macOS shortcut is derived from the official guide's own PC to Mac modifier table, which maps Ctrl to Command and Alt to Option, giving Cmd+Option+Z, and the menu location is stated plainly because the entry sits under Edit rather than the View menu where several other programs put a history panel. The window's right-click menu is covered for what it offers beyond jumping, namely docking or undocking in the docker and removing selected states from the history, with the caution that removing states throws away routes back and is housekeeping rather than a recovery move. The feature the article argues is worth learning the window for is the option to store multiple redo paths where possible: with it enabled, returning to an earlier point and then doing something new does not necessarily destroy what came after the original point, because REAPER stores the new sequence as an alternate set and remembers both paths independently, creating another path each time that point is revisited. The branch count is visible in the list, so an entry flagged with a count such as (*2) carries that many additional paths beyond the original sequence, and right-clicking there chooses which state to load. Persistence is treated as the part that actually decides whether any of this helps tomorrow. Two settings on the Options, Preferences, General, Undo page do the work, Save undo history with project files and the matching Allow load of undo history, and the official guide states that with both enabled the history is available next time the project is used even after REAPER has been closed and restarted. A check that needs no menus is given alongside it: saving the history produces a separate file beside the project with the extension RPP-UNDO, documented in the guide's project file types table as the Undo History File and created only when that option is on, so a project folder with no such file beside the .RPP settles the question. The point is made that enabling these options does nothing for a project already closed without them, because the history has to have been written at save time, which makes this a setting to change on a quiet day rather than a recovery tool. The remaining undo preferences are tabulated rather than prescribed, deliberately without claiming defaults, covering the maximum memory allocated to undo, retaining the most recent actions when the undo area is full, whether item, track, envelope point, time selection and cursor position changes create undo points at all, the two persistence options, multiple redo paths, and showing the last undo point in the menu bar so the reader can see what Ctrl+Z is about to do before pressing it. The limits section is explicit that this is a record of edits and not a backup: it will not bring back a media file deleted or moved on disk, which is an offline-media fault with a different fix; it will not recover a history that was never written, because without the save option there is no file and no setting changed afterwards retrieves it; it will not contain changes the preferences were told not to record; it will not help when the project file itself is damaged or REAPER went down mid-session, which is a recovery job; and it is not version control, since a saved history is one file beside one project and is overwritten as work continues. A five-step list closes it for the case where the wanted state is not in the window, starting with checking which project tab is in front because undo history is per project, then looking for the RPP-UNDO file, then checking whether the action creates an undo point at all, then considering that the memory ceiling was reached and the oldest states went, and finally noting that for a single word or phrase, punching in is often faster than hunting backwards through the history. Every claim is sourced to the official REAPER user guide v7.81 and the official changelog, which was checked on the day of publication against the current release v7.82 of 4 October 2026. No testing, timings or defaults are claimed, and no link is an affiliate link.
Podcast and Spoken-Word AudioHow to export a podcast episode as a video for YouTube from REAPER
The article treats putting a podcast on YouTube as a delivery problem rather than an audio one, and its first move is to establish that the obstacle is real and documented: YouTube's own supported-formats page states that audio files such as MP3, WAV or PCM cannot be uploaded to create a YouTube video and points the reader at editing software, so an audio-first show has to be wrapped in a video before it can go up. Before any render settings are given, the reader is sent to check whether the render is needed at all, because YouTube's RSS ingestion route will create a static-image video from the show art and upload it on the creator's behalf. That route is given with its published limits rather than as a recommendation: it is available in select countries and regions only, the content cannot contain advertisements, episodes arrive private, the result appears on YouTube and YouTube Music only, every episode carries the same show art, and audio cannot be updated once published, so a correction means re-ingesting the episode, which creates a new video and makes the old one private. Rendering from REAPER is positioned as the answer when the reader wants a per-episode image, a title or graphic on screen, a file for somewhere other than YouTube, or has no feed at all. The REAPER path is then given in order: finish the audio first, because the video wrapper changes nothing about how the episode sounds; prepare a 1920 x 1080 image to match the render and YouTube's 16:9 desktop standard; drag the image in as a still-image media item and stretch it across the whole episode, noting that REAPER's default length for an imported still image is a preference rather than anything meaningful; and confirm it is on screen with View, Video Window, Ctrl Shift V. Two practical traps are named from the official user guide, that where two tracks hold video or image items only the higher one is displayed, and that the stock video processor's Resize track/item to project dimensions preset is the clean fix for an image whose proportions do not match the render size. The render targets are tabulated against YouTube's published recommendations: MP4 container, H.264 High Profile video, AAC-LC audio at 48 kHz and 384 kbps stereo, 1920 x 1080, and the lowest frame rate REAPER offers, which the guide documents as 24, on the argument that a still image needs none of them. Two naming changes are flagged because they break existing tutorials: REAPER 7.70 renamed the video encoder format dropdown to file, and 7.66 added separate MOV and MP4 FFmpeg modes while deprecating the old combined QT/MOV/MP4 auto-select entry that older guides and the current printed user guide still describe. The codec-availability problem is treated as a build limitation rather than user error, with the user guide's own remedies: the Windows build ships a limited FFmpeg and the full libraries' DLL files go in the UserPlugins folder found via Options, Show REAPER resource path in Explorer; macOS AVFoundation writes .m4v and .mov only, which is not a dead end because MOV is on YouTube's supported list; and WebM with VP8 and Vorbis is an accepted fallback nobody should prefer. REAPER's native Windows Media Foundation MPEG-4 encoding is noted from the official changelog as a second Windows route, with the honest caveat that what appears in the Output format dropdown depends on platform and build and that the reader should read their own dialog rather than hunt for a name from a tutorial. MKV is called out as a container REAPER offers that is not on YouTube's supported list, and the None codec option is noted as the way the same dialog produces an audio-only or video-only render. An optional section covers the stock video processor presets that go beyond a flat image, Image overlay and Text under Overlay and the Decorative Oscilloscope with Blitter and Decorative Spectrum Analyzer under Synthesis, with Cockos' own word decorative taken at face value: they are visuals rather than meters, and they turn a cheap still-image encode into a real one. The limits section states that the wrapper will not improve the audio, will not change playback loudness, will not produce chapters, will not make YouTube treat the upload as a podcast, and will not sharpen a small cover image scaled up. A four-point pre-upload check closes it: play the end of the file for an image item that stopped short, confirm an audio codec was not left on None, confirm the file reports 1920 x 1080, and confirm the render bounds were not left on a time selection. No listening tests, measurements, render times or file sizes are claimed, every setting is given as a target rather than a guarantee, and no link is an affiliate link.
Home Vocal Recording and MixingCan this vocal take be fixed in the mix, or do you need to re-record it?
The article is a triage rather than a mixing tutorial, and its position is stated in the first paragraph: most takes that sound bad are fixable, because the usual complaints are tone and level problems that survive processing intact, and only four kinds of damage are genuinely unfixable. Before any of the four is assessed, the reader is sent to check one thing that can reverse the verdict, which is whether the file is 32-bit float. The argument is that a flat-topped waveform means different things in different formats: in a 24-bit or 16-bit file full scale is a hard ceiling and anything past it was never captured, whereas a 32-bit floating-point file can represent values above 0 dBFS, so overload that happened in the digital domain after the converter may still be sitting in the file waiting to be turned down. RODE's own REAPER setup guide for 32-bit float recording is cited for both the path and the claim, setting the WAV bit depth to 32 bit FP under File, Project Settings, Media, Recording, recovering a hot take through Item Properties, Take Properties, Volume/pan, Normalize with the peak target lowered from 0.0 dB to around -3.0 dB, and stating that after levelling the previously clipped audio information is still intact. The honest limit is given immediately afterwards and is the part most marketing blurs: the format protects against overload after the analogue-to-digital converter and does nothing about overload before it. Sound on Sound's review of the RODE NT1 5th Gen is used for the mechanism, that the microphone's clipping resistance comes from two parallel A-D converters with one aligned so the analogue circuitry cannot clip even at maximum gain, that the native digital output is a 32-bit floating-point signal, and that selecting the mic as a 24-bit source with the gain too high results in a clipped recording. The four unfixable categories are then given with a test and a verdict each. Real clipping and preamp overload, where the Audacity manual's documentation for Clip Fix is quoted for the limits of the whole category, that it attempts to reconstruct clipped regions by interpolating the lost signal and is only likely to be effective for repairing lightly clipped audio; the test is how long the flat sections last and whether crunch appears on every loud word. A noise floor louder than the quiet parts, where the mechanism is that gain is linear so the ratio between voice and noise is fixed at capture, supported by Neumann's calculation of signal-to-noise as 94 dB minus self-noise in dB-A and their observation that even a very quiet recording room contributes quite a bit more ambient noise than 10 dB-A; the test is to raise the quietest line the reader intends to keep rather than the loudest. Room sound printed onto the take, where the mechanism is that reflections reached the capsule summed into the same waveform and REAPER ships no stock tool that separates them, and where the article adds the twist that re-recording only helps if the room is actually changed. And pitch or timing a long way out, sourced to the official Cockos documentation for ReaTune, including automatic correction against Key, Mode and an optional Notes Allowed list, the Attack Time control described as the time between the start of a note and when ReaTune starts to correct it, and the Prevent octave shifts option that exists because harmonics can make the detector guess the wrong octave, plus the play rate control with Preserve pitch enabled using the SoundTouch time-stretching algorithm. A counter-list follows of the things that sound unfixable and are not, covering muddy, boxy, nasal, thin and harsh tone, piercing sibilance and thumping plosives, wildly uneven levels, a single wrong word that can be punched in rather than re-recorded, and the case where several passes exist and comping beats both fixing and re-recording. The perception case is separated out as not a recording problem at all. A six-step five-minute triage is then given in order, and the article notes that if none of the four categories fired, the reader is listening to an unmixed vocal rather than a broken one. The limits section states that the triage will not judge whether the performance is any good, that there is no numeric threshold because usability depends on the density of the arrangement, and that a take failing two categories at once is almost always a re-record. The closing section argues that re-recording without changing the level, the distance to the microphone or the space produces the same recording with a different performance on it, and advises muting rather than deleting the original take. No listening tests or measurements are claimed, all processing figures are given as starting points, and no link is an affiliate link.
Audio Repair and RestorationRecorded with noise suppression on? What can and cannot be fixed in REAPER
Most writing about background noise answers the opposite question, and the article's position is that a recording with chopped word starts and dead gaps was damaged before REAPER ever saw it, by noise suppression or a gate running upstream, and that this damage is not reversible. The symptom list is given first so a reader can tell the two problems apart: the first syllable after a pause going missing because the detector spends it deciding that speech has begun, quiet tails and breaths vanishing, gaps that are a flat line at zero rather than a thin band of room tone, a background that pumps with the voice, a swirly warble on quiet consonants, and non-vocal material suffering worst, which RODE's own support article supports by noting that Windows 11's audio enhancements can leave non-vocal audio muffled or in some cases silent. Three layers are then named as the possible source. The operating-system layer is sourced to Microsoft's driver documentation on audio processing objects, which defines them as customisable software based digital signal processing for Windows audio streams, states that the technology is symmetric and works essentially in reverse for input devices, and says an endpoint effect is always applied even to raw streams; RODE adds that the Windows audio enhancements setting is applied per device, so switching it off for one microphone does not cover another. The macOS equivalent is Mic Mode, with Apple's own instruction to select Voice Isolation to filter out background sounds, and Focusrite's support article quoted directly for what that costs a recording made through an interface: it can alter the audio quality of your input and your channel mapping, it only works for speech, and unless specifically required they recommend disabling it. The remaining two layers are the microphone's companion software and the app the take was recorded in, and the article warns that all three stack. A two-minute diagnostic test follows, recording ten seconds of silence and then counting to five with clear pauses, then reading the waveform for a dead flat gap and for square word attacks. The irreversibility argument is given as mechanism rather than assertion: suppression decides moment by moment and band by band how much of the signal is noise and reduces those parts, the file stores only the result, the per-moment gains are stored nowhere, so there is nothing to invert, and a gate that closed wrote silence which contains no information about what it replaced. Four things that genuinely help are then given in order, each framed as improving how the damage reads to a listener rather than recovering content: laying surviving or newly recorded room tone under the take so the pumping stops drawing attention, softening square word boundaries with REAPER's item fade handles using Alt left click to delete a fade and Ctrl left click to cycle shapes with the global behaviour under Options, Preferences, Project, Item Fade Defaults, rebuilding by hand only the words that actually break a sentence by punching in or editing in a pickup line, and riding levels with a volume envelope rather than compressing, because compression lifts the pumping and the artefacts along with everything else. One optional high shelf is offered at a starting range of +1 to +3 dB above roughly 6 kHz, with the explicit trade that the same shelf lifts the warble. What to avoid is given as four don'ts: no further noise reduction on a file with no stable noise floor to learn from, no second gate, no heavy limiting, and no expectation that a paid restoration tool changes the ceiling, because it is still smoothing rather than restoring. The pre-flight check is presented as the part worth keeping, four ordered steps across Windows audio enhancements set per input device, macOS Mic Mode set to Standard, the microphone's companion app, and the recording app, followed by re-running the two-minute test to confirm the gaps have room tone in them. The limits section states plainly that missing syllables were never written to the file, that a substitute room tone bed is a convincing mask and not the original, that voice-focused suppression usually leaves music and instruments beyond useful repair, and that a one-take interview with lost words needs a pickup line or a transcript correction rather than a plugin. No listening tests or measurements are claimed and no link is an affiliate link.
REAPER Audio WorkflowsHow to use REAPER's Media Explorer to audition and import beats and samples
Most people meet REAPER's Media Explorer, decide it looks like a worse file browser and go back to dragging audio in from a folder window, and the article's position is that this costs them the one thing the window exists for, which is hearing a file properly before it becomes part of a project. Every mechanism is taken from chapter 4 of the official REAPER User Guide v7.81, read on October 18, 2026: the window opens with View, Media Explorer or Ctrl Alt X and is dockable, its toolbar runs Insert item in project, Auto play, Start on bar, Detect pitch and the media information toggle followed by the Tempo match options, a Preserve pitch toggle and the dock toggle, and folders created inside it are real folders on disk while databases exist only inside the Media Explorer. Auditioning is given as a keyboard workflow, selecting a folder or database in the left panel and moving through the file list with the up and down arrow keys while Auto play starts each file, with the right arrow or spacebar to play and the left arrow or spacebar to stop, the repeat toggle to loop, a time selection dragged in the preview window to hear only part of a file, the Options menu's Fade in Fade Out Settings to stop preview clicks, and Auto advance to next file after preview to walk a folder hands free. The clearest unique contribution is the Route button, which directs preview output to any audio output or through a track selected in the arrange view, so inserting one track and keeping it selected turns the Media Explorer into a listening station that plays samples through the reader's own volume, panning and FX rather than dry. Tempo match is then explained as a mechanism rather than a button: with it on the selected item's tempo is matched to the project, half and double options exist for loops that read at the wrong multiple, and the reason it appears dead is that matching needs tempo information attached to the file, which the official 7.79 changelog confirms by stating that media is stretched automatically on import from the Media Explorer when tempo match is enabled and the file contains tempo information. That version change is flagged as a behaviour change rather than user error, and the recurrence of the confusion is evidenced by two Cockos forum threads read on the day, 186758 from January 2017 about MIDI files and 293195 from July 2024 about loops failing to sync during preview, cited as evidence that the question recurs rather than as documentation. The five insert commands are listed exactly as the manual gives them, insert into project on the selected track at the cursor, insert at time selection stretching or looping to fit, insert on a new track, insert into sample player, and insert in selected item as a new take or replacing the media source, with Options, Default action setting which of them a double click or Enter performs and the Media Explorer's own action list section available from the title bar for shortcuts. A separate note covers replacing one sample across a whole song by holding Ctrl Alt while dragging onto the target item. What carries over from the preview is treated as a trap in both directions: pitch, rate and volume set while auditioning are preview settings only unless the matching apply options are enabled, so some readers are surprised a sample does not arrive at the level they heard and others forget a nudged pitch knob, and the recommendation is to leave those off and decide level and pitch on the timeline. Shortcuts and databases are covered briefly, Add to shortcut list to pin a sample drive, Make database from folder which appears prefixed DB: and can be scanned for new files, and the filter box with Options, Search choosing searchable fields. The limits section states that the Media Explorer will not improve the audio, will not find a tempo that is not in the file, will not make stretched audio sound unstretched, is not a tagging library with ratings and similarity search, and will not reconnect media a project has already lost. A six step troubleshooting order ends on the two settings that most often explain a wrong level or an unwanted fade, the apply-preview options and Preferences, Project, Item Fade Defaults. No listening tests or measurements are claimed, and the current release is noted as 7.82, dated October 4, 2026, against a user guide at 7.81.
Tools and RecommendationsREAPER vs Audacity for podcasting, and what Audacity 4 changed
Most REAPER against Audacity comparisons were written against Audacity 3 and have not been revisited, and the article's position is that this, rather than any change of character in either program, is why the question got loud in September 2026. The dated facts come from Audacity's own 4.0.0 release notes and releases list, retrieved October 17, 2026: Audacity 4.0.0 was published on 3 September and is marked Latest, Audacity 3.7.9 was published on 1 September, and releases back to 3.0.0 remain browsable. What 4.0 adds is set out first and fairly, a rebuilt interface with high-DPI rendering, dockable panels, saved workspace layouts, a home screen, direct clip selection, multi-clip editing, clip grouping, a dedicated split tool, a draggable playhead, seeking without stopping playback, recording from any point on the timeline, ASIO in the official Windows builds and meters in the track headers. The release notes' own not-yet-available list is then given in full, because it is the entire reason anyone is asking: Macro Manager and the scripting pipe, the Mixer, Play-at-speed, VAMP and LADSPA plugin hosting, Time Tracks and Note or MIDI tracks, plus some exporting and rendering features, analyzers and effects, with supported plugin formats narrowed to VST3, Nyquist, LV2 on Linux and Audio Units on macOS. Two items were removed rather than deferred, the separate Select, Envelope, Draw and Multi-tool modes and Sync-Lock, which the transition guide says ripple editing replaces. One actionable warning is pulled out rather than buried: Audacity 4 uses a new .aup4 project format, an .aup3 project converts without altering the original, but a converted project cannot be saved back to .aup3, which matters to anyone sharing project files with a collaborator still on Audacity 3. The article's central move is to refuse a single verdict and split the reader into three groups with three different answers, carried in a table. Someone recording one voice, editing by hand and exporting an MP3 should stay on Audacity 4, because they never used the missing features and it remains free. Someone who lost a Macro, a scripted batch step, Play-at-speed or a LADSPA plugin should go back to Audacity 3.7.9, which is still offered for download, and wait, and this is named as the cheapest and most likely correct answer rather than passed over in favour of the paid option. Only someone recording co-hosts or guests on separate tracks, repeating the same chain every episode, or processing folders of files is pointed at REAPER. The REAPER case is built from the official Quick Start guide: per-track FX chains that reorder by dragging and can be saved and loaded from the FX chain window, track templates saved and inserted from the Track menu, track inputs assigned by right-clicking the record arm button so two mics stay independent, and rendering to a master mix or to stems. Cost is quoted exactly from the Cockos purchase page as read on the day, 60 dollars for the discounted license against 225 for the commercial one, the discounted eligibility wording in full, 60 days of evaluation free with full functionality and no strings attached, and free upgrades through version 8.99. The limits section is the part the comparison pages omit: switching editor will not make a recording sound better, will not rescue a bad room or a clipped take, will not restore Audacity macros because REAPER's batch converter works on files while repeating edits inside a session is a different mechanism built from custom actions, will not be faster in the first week, and should not be assumed to fix an accessibility problem in either direction. Three setup steps close it, confirming plugins scan, saving the first session as a track template, and choosing a loudness target before rendering. On Macros there is no published return date: an Audacity contributor stated on 10 September 2026 that they are not in 4.0 and may be added to a future version, the tracking issue is GitHub 9407, and a forum user's suggestion of 4.1 is labelled unsourced rather than repeated as a roadmap. No listening tests, benchmarks or installations are claimed, no link is an affiliate link and neither developer paid for placement.
Practical MixingHow to use REAPER's stock limiter without squashing your track
Home producers reach for a limiter because their render sounds quieter and smaller than released music, and the article's position is that the squashing that follows is a predictable consequence of one control being misunderstood. REAPER's stock brickwall limiter is ReaLimit, and its mechanism is taken verbatim from section 16.14 of the official REAPER User Guide v7.81: it lets you set a threshold and a ceiling and then transfers sound from the threshold to the ceiling, so sound arriving at the threshold leaves at the ceiling, nothing can leave above the ceiling, and everything above the threshold is compressed. The manual's own worked example is carried through, a threshold of -6 dB against a ceiling of 0 dB boosting the whole signal by 6 dB, which establishes the key point the article is built on: the threshold control is a loudness control, not a safety control, and that is the entire squashing mechanism in one sentence. Setup is given in four ordered steps, ReaLimit last in the master FX chain so nothing downstream can push peaks back over the ceiling, the ceiling set once at about -1.0 dB and then left alone, the threshold lowered only while the loudest section plays until a few dB of reduction shows on the biggest moments and none shows in the quiet parts, and a bypass comparison after every move. The ceiling is argued at -1.0 dB rather than 0 dB on Spotify's own published guidance, which asks for true peak below -1 dB and below -2 dB for masters louder than the normalization target because louder tracks distort more readily when encoded. One correction is the article's clearest unique contribution: ReaLimit's release is expressed in dB per second, not milliseconds as most writing about it assumes, and the manual's two failure modes are given as stated, audible pumping when it is overly long and distortion artefacts when it is too short, with the advice to leave it alone until a specific problem has actually been heard. A seven row table covers every remaining control with the manual's own description against a practical starting point, including Constant gain managing threshold and ceiling relative to each other, True peak as the more accurate reading, Performance offering high quality by default or low latency, and the note that the Metering selection affects the display only and not the processing. The stereo, multi-mono and multi-stereo input modes are distinguished, with stereo recommended for an ordinary song or voice track so the two sides stay together, and the manual's own observation is carried that ReaLimit is not confined to the master and is often more useful gluing a folder or submix. On how much reduction is too much the article declines to give a universal number and gives a method instead, single digits on the loudest moments with the meter resting at zero through quieter sections, and names the honest read when heavy limiting is needed, which is that the limiter is being asked to fix the mix. The loudness chase is then argued to be largely self-defeating, because Spotify normalizes playback to -14 dB LUFS under the ITU 1770 standard and applies negative gain to louder masters, so a crushed track is turned back down and what survives is the dynamic range it was paid for with. The limits section states that a limiter will not fix an unbalanced mix, will not undo clipping already in the file, will not add width or depth, will not make a thin mix full, and will not protect a track through encoding if the ceiling sits at 0 dB. A five item troubleshooting order ends on comparing at matched loudness before concluding there is a problem at all. No listening tests or measurements are claimed and all settings are given as starting points. Both sources were retrieved and checked on October 16, 2026.
Home Vocal Recording and MixingWhat sample rate and bit depth should you record vocals at in REAPER
The question has a boring answer and the article's position is that saying so plainly is the useful contribution, because sample rate is one of the smallest factors in how a home vocal sounds while the room, mic position and record level all matter more. Bit depth is separated out as the part with a real answer: record 24-bit, for headroom rather than audible silence. The reasoning is taken from Analog Devices' MT-229 tutorial, which gives an ideal N-bit converter's signal-to-noise ratio as 6.02N + 1.76 dB, putting 16-bit near 98 dB and 24-bit near 146 dB, and the honest part is labelled immediately: no real interface reaches the 24-bit figure, so the extra bits buy permission to track conservatively with peaks around -18 to -12 dBFS rather than inaudible noise. Sample rate is then decided on compatibility rather than quality, with each branch tied to a published specification retrieved the same day. 48 kHz is the pick when video is anywhere downstream, because Google's recommended upload encoding settings for YouTube specify 48 kHz audio and video timelines already run at that rate. 44.1 kHz is the pick for music streaming and audiobooks: Spotify's delivery requirements accept 44.1 kHz or higher, want 24-bit where that is the native master and 16-bit only where no higher-depth master exists, and reduce anything above 24-bit to a maximum of 44.1 kHz and 24-bit; ACX is the one hard requirement on the page, since every delivered file must be a 192 kbps or higher constant bit rate MP3 at 44.1 kHz. 96 kHz is argued against for a single voice on cost grounds, doubling disk use and roughly doubling per-plugin load in exchange for content above 24 kHz that nobody hears and that will be converted down on the way out, with a callout making the fair case that pitch shifting, extreme time stretching and analogue modelling are the real reasons higher rates exist. 32-bit float gets its own answer: it is for the unrepeatable take at a level you cannot control, not the default, and Spotify asks for 24-bit anyway. A five row table maps delivery target to rate and depth. The REAPER section is three ordered steps in the order that matters, the audio device first, then the project sample rate, then the Media page record format, with REAPER's WAV options named as 16-bit, 24-bit and 32-bit float and an instruction to read your own dialog rather than trust a stated default, because the Cockos wiki page carrying REAPER's 44.1 kHz and 24-bit default is flagged as needing an update. The limits section is the part most versions of this article omit: these settings will not improve a room, will not make a cheap mic sound expensive, will not undo clipping that already happened, will not fix muddy, nasal or harsh tone, and will not survive an interface and a project that disagree. A four item troubleshooting order ends on the distinction between a recording specification and a delivery specification, which is usually the one that actually failed. All platform specifications were retrieved and checked on October 15, 2026.
Tools and RecommendationsThe best free pitch correction plugins for vocals (and when ReaTune is enough)
Free autotune roundups are almost always written as a ranked list of nine downloads, and the article's position is that this is the wrong shape for a REAPER user, because one of the candidates is already installed and the real question is what a download would add to it. The stock baseline is set first from Cockos's own effects guide: ReaTune ships with REAPER, currently version 7.82 released October 4, 2026, and carries three tabs, a tuning meter, automatic correction and manual correction. Automatic mode exposes a key name from A to G sharp, a key mode or scale whose default is Chromatic, individual on and off switches for all twelve notes, an attack time governing how long after a note starts correction begins, and a choice of pitch-shifting algorithm including three elastique 2.1 variants; manual mode draws detected pitch as a graph so correction bars can be placed only where they are needed. The Chromatic default is named as the single most common reason people conclude ReaTune sounds bad, because it snaps to the nearest of all twelve notes and can lock a note that sat between two pitches onto the wrong one. Criteria are fixed before any pick is made: genuinely free with no time limit, runs as a plugin in REAPER on Windows or macOS, does something stock does not do easily, is honest about what sits behind the paywall, and leaves a one-click bypass, because heavy correction is very easy to stop hearing. Three downloads clear that bar. Auburn Sounds Graillon 3, version 3.2, offers a Free Edition carrying the pitch shifter, formant shifting, pitch correction in natural and hard-tune forms, three pitch engines and built-in compressor, gate, chorus, preamp and bitcrusher, as VST, AAX, AU and LV2 on Windows, macOS and Linux; the Full Edition at $29 adds pitch-tracking modulation, keyboard tuning for non-equal temperaments and a vocal doubler, and the doubler is identified as the omission most home vocalists would actually notice. MeldaProduction MAutoPitch is free automatic tuning with formant shifting and stereo expansion, delivered inside the 38-effect MFreeFXBundle, and two caveats are taken from Melda's own pages rather than softened: you install a bundle to get one tuner, and since version 17.00 installation runs through MPluginManager, which signs into a Melda account to check licenses and activate plugins, so this is a free plugin with an account step. Session Loops PitchNet is the pick the generic lists do not carry, a free and open-source pitch editor under the GNU Affero General Public License v3 that replaces the correction curve with a piano roll, detects pitch with neural models, resynthesises edits through a neural vocoder and exports WAV or MIDI, with prebuilt free installers for Windows (VST3, AAX) and macOS (VST3, AU, AAX universal) and published requirements of Windows 10 64-bit or macOS 13 or later, 8GB of RAM with 16GB suggested, about 500MB of disk and a suggested dedicated GPU. Two honest labels are attached to it: the community enthusiasm includes a user reporting it beat Melodyne on difficult correction, which is one subjective impression posted in a thread started by the developer rather than a measured result, and the GPU suggestion makes it the wrong tool on a laptop already short of headroom. GSnap is addressed rather than ranked, still free and still downloadable from GVST but an older VST-format plugin, which matters because a current Mac user is looking for VST3 or AU and a plugin that never appears in the FX browser is usually a scan path or format problem. A six row table maps the reader's situation to a pick, including two rows whose answer is the plugin already installed. The limits section is the part the ranked lists omit: pitch correction does not fix timing, does not rescue an uncommitted take, cannot track a stacked or wet vocal because every tool here detects one pitch at a time, does not fix tone, is not invisible when pushed, and matters less than the gap between a rushed take and a warmed-up one. A five item troubleshooting order ends on bypassing the plugin and listening to the raw take again, because ears normalise to correction within a few passes. No listening tests are claimed, no plugin developer pays for placement and no link in the article is an affiliate link. All product pages were retrieved and checked on October 14, 2026.
Home Vocal Recording and MixingHow to record vocals over a mastered beat that is too loud in REAPER
A beat bought or streamed as a finished master is loud by design, and the article's position is that the resulting crunch is an arithmetic problem at the master output rather than anything wrong with the voice, the microphone or the recording level. Mastering compresses and limits a file so its loudest peaks sit just under 0 dBFS, which is not a guideline but the top of a fixed point scale: a 16 or 24 bit WAV has no value available above it and neither does an interface output. A vocal is a second signal summed onto that, so the sum exceeds the ceiling and the excess is flattened, which is why the vocal track's own meter can read healthy while the mix sounds damaged, and why raising the vocal or compressing it harder makes the result worse rather than better. The counterintuitive part of the argument is that nothing is lost by turning the beat down, because playback loudness is normalized downstream: Spotify publishes a target of -14 dB LUFS based on the ITU 1770 standard and applies negative gain to anything louder, and the measurement itself is defined by Recommendation ITU-R BS.1770-5 (11/2023), “Algorithms to measure audio programme loudness and true-peak audio level”. A loud master is therefore turned back down on playback whether or not the session was loud. Diagnosis is three checks that each decide something different: soloing the beat and reading its track meter, which identifies a mastered file when it sits within a couple of dB of 0; watching the master rather than the vocal with both playing, which locates the problem in the sum; and zooming in on the item, where a limited master reads as a near solid block with flat tops. A fourth case is separated out, the beat that pushes the master into the red on its own, which is damage printed into the file and a repair question rather than a level question. The fix is a single decision, how far down the beat goes, given as a starting range of 10 to 14 dB to be checked against the master meter rather than as a correct number, with the instruction to record what was used so the next session starts in the same place. Three places to make that cut are distinguished rather than treated as interchangeable: the beat track's volume fader, which is simplest for a single stereo file and survives swapping the beat; item volume via item properties, which travels with the media item and is therefore easy to forget; and a folder track used as a beat bus when the beat arrived as stems, which moves the whole instrumental while preserving the internal balance the producer intended, citing Cockos's own description of REAPER's nested folder system as handling “routing, bussing, all in one step”. Pulling the master fader down instead is rejected explicitly, on the grounds that it stops the overs without changing the beat to vocal relationship and leaves every later level decision being made into a master that is not at unity. The tracking order that follows is deliberately short because two of its three steps are answered elsewhere on the site: set the record level once and leave it, build the headphone mix separately because the beat usually needs to be louder in the phones than in the mix, then judge the balance on the loudest chorus rather than a verse and take another 2 to 3 dB off the beat rather than off the voice if the master still touches red. The limits section is the honest core. Lowering a mastered beat restores headroom and not dynamics, because the limiting is printed into the file, so the beat will sound equally relentless and merely quieter; the win is that the voice sits on top of it, not that the beat opens up. It will not make the vocal sit in the mix, which is a separate balance and EQ question; it will not rescue a take recorded too quietly or too hot, since beat level and input level are independent; and it will not change final upload loudness, which normalization sets. Readers are advised to ask the producer for an unmastered or track-out version with headroom left in it before working around the mastered one. A five row table maps symptom to cause, test and fix, including the stems case where each file was lowered by a different amount. A five item troubleshooting list ends on the distinction that matters: if the crunch survives a master meter that is comfortably clear, the damage is in the file or the take and not in the arithmetic. Every source was retrieved and checked on October 13, 2026.
Podcast and Spoken-Word AudioHow to add chapters to a podcast episode (and override the automatic ones)
Chapters are the podcast feature whose instructions are almost always written by a hosting company about its own dashboard, and the article's position is that the confusing symptoms downstream all come from three different mechanisms being treated as one thing. The first is a list of timestamps and titles in the episode description. The second is chapters supplied through the host so they travel in the RSS feed, which Apple documents as the <podcast:chapters> tag, whose two required attributes are a url pointing at an external chapters file and a type whose preferred value is application/json+chapters, and whose argument is that the chapters live outside the audio and can be changed after publishing. The third is chapters written into the audio file itself, defined by the ID3v2 Chapter Frame Addendum: a CHAP frame per chapter carrying start and end times in milliseconds from the beginning of the file, optional byte offsets and optional sub-frames for name and description, plus a CTOC frame for the table of contents. That document is noted as an informal standard dated December 2005 rather than a formal part of ID3v2, which the article gives as a large part of why app support varies. The platform comparison is the practical core. Apple Podcasts documents all three routes; Spotify's creator documentation describes only description timestamps and, for shows hosted elsewhere, Podlove chapters in the RSS feed, and does not mention embedded ID3 chapters or the podcast:chapters tag at all. The description list is therefore the only single route both of the two largest platforms document, and the same list satisfies YouTube, whose published requirements are a first timestamp of 00:00, at least three timestamps in ascending order and a minimum chapter length of 10 seconds. The newest part of the question is automatic generation. Apple generates chapters for full and bonus episodes in English starting with iOS 26.2, labels them Automatically Created, excludes trailers and episodes under 10 minutes, and generates none at all where it cannot produce a transcript; chapters the creator supplies are always displayed instead, including on already-published episodes, the automatic list can be downloaded from Apple Podcasts Connect as a TXT file and edited, and the feature can be switched off per show on the Availability tab or per episode. Apple's published guidance is given as numbers: at least three chapters, chapters for episodes longer than 10 minutes, nothing shorter than two minutes, no more than six per hour. Spotify also generates chapters from a transcript, in English only and not yet for all shows, and states that manual chapters take priority. The REAPER half is the editing habit that makes the timestamps free: named markers dropped while the session is open, using the shortcuts in the Cockos Quick Start guide, placed on the first word rather than the breath, with a warning that timestamps must be taken from the locked edit because a later cut invalidates everything below it. A worked four-chapter list is shown. The limits section is explicit that chapters are navigation and not a growth strategy, that they do not touch the audio, that they will not look identical across apps, that they do not survive a re-edit, that one host documents chapter times being adjusted to match the current campaign under dynamic ad insertion, and that audiobook chapters are a different delivery model entirely. A seven-item troubleshooting list ends on the most common cause, which is a first timestamp that is not zero. All sources were retrieved and checked on October 12, 2026.
Audio Repair and RestorationRecorded on the laptop mic instead of your interface: what can be saved
A take that went to the laptop's built-in microphone instead of the interface is usually partly salvageable for speech and effectively unsalvageable as a music vocal, and the article's position is that this is not simply a worse-microphone problem but three separate problems that have to be told apart before any plugin goes on the track. The first is the capsule: published figures for one widely used MEMS part, the Knowles SPH0645LM4H-B, give a signal to noise ratio of 65 dB(A) and an acoustic overload point of 120 dB SPL, against Audio-Technica's published 74 dB signal to noise ratio and 20 dB SPL noise figure for the AT2020, roughly 9 dB of difference, disclosed as an indication rather than a measured comparison because the Knowles figure is A-weighted and Audio-Technica prints no weighting, and labelled representative of the class rather than a measurement of any particular laptop. The second is placement, which the article treats as the larger factor: the port sits in the chassis near the hinge, keyboard or fan and two to three feet from the mouth rather than six to eight inches, which is why the take is roomy and why key clicks and fan noise are conducted straight into the capsule. The third is the operating system, and it is what limits the repair. Microsoft's driver documentation is cited for the processing that can be applied to a capture stream, including Simple Noise Suppression, which "suppresses noise such as humming and buzzing", Automatic Gain Control, which is "designed to provide a controlled signal amplitude at its output, despite variation of the amplitude in the input signal", and Deep Noise Suppression, which "uses advanced AI / machine learning techniques to suppress noise particularly in voice calls". Two statements from that same page decide what the reader can do next: "OEMs define what effects will be used for each mode", so the behaviour is a vendor decision rather than a Windows standard, and applications "cannot find out what mode is used for each of their streams", so REAPER cannot report what was done and neither can the user. Only raw capture is guaranteed clean, and Microsoft defines it strictly as having to exclude "any time varying or adaptive processing, such as echo control, automatic gain control, or noise suppression". Apple's macOS mic modes are given as the Mac equivalent, with Voice Isolation to "filter out background sounds" and Wide Spectrum to "include the sounds around you", and the note that options vary by app. The consequence is stated plainly: adaptive processing changes moment to moment, so once it is printed there is no inverse to apply. Three diagnostic checks come before processing, each deciding something different: whether the noise floor swells during long pauses, which identifies gain control; where the top end stops when watching ReaFIR's live spectrum display on a loud bright word, a cliff meaning the content is simply absent; and whether the residual noise is steady hiss, which is repairable, or warbling and watery, which is already a processing artefact that further reduction makes worse. The repair order then runs high pass around 90 to 130 Hz, higher than usual because little useful low end was captured to protect, a swept band cut of roughly 2 to 5 dB somewhere in 300 to 700 Hz for the boxiness of a distant capsule in an untreated room, a high shelf of about 1 to 4 dB in 3 to 6 kHz but only where check two showed real content to lift, a single light pass of ReaFIR Subtract for steady noise following the Cockos effects guide's own instructions to enable "Automatically build noise profile" "while auditioning a section which contains only the noise" and then "un-check Automatically build noise profile, stop playback and save the preset", with its warning that when the profile is raised "the higher you raise it, the more artifacts you will hear", and levelling last rather than first because EQ changes how loud the file reads. Every figure is given as a starting range to adjust by ear. The limits section is explicit that missing high frequencies cannot be restored and anything claiming to is generating new material, that automatic gain control cannot be undone because the original dynamic shape is gone, that a distant capsule prints room reflections that de-reverb only partly masks, that keyboard and fan noise overlapping a word stays, and that the target is clear and listenable rather than transparent. A five row table maps what the reader hears to cause, action and a realistic outcome, including two rows whose honest outcome is that no improvement is available. The article opens and closes on the same decision rule: if the performance can be repeated, repeating it beats every repair below and takes less time, so the whole method is addressed only to takes that genuinely cannot be repeated, and a laptop take intended for a music mix is treated as a scratch reference rather than a candidate for repair. Prevention is given as two habits, tapping the intended capsule and watching which meter moves, and setting the input deliberately rather than inheriting the last assignment. No listening tests are claimed and no specific laptop model is identified as applying any specific processing, because vendors do not publish it. All sources were retrieved and checked on October 11, 2026.
Podcast and Spoken-Word AudioHow to clean up background noise that changes through an episode in REAPER
Noise reduction that worked perfectly on the first minute of an episode and then stopped working is not a settings mistake, and the article's position is that stationary and non-stationary noise are two different jobs that the usual advice treats as one. ReaFIR's Subtract mode captures a fixed picture of the noise and subtracts that same picture everywhere, a method the Cockos effects guide describes as enabling "Automatically build noise profile" over a noise-only section and then unchecking it once satisfied, so the accuracy of the result depends entirely on the stored profile still matching the noise present. Three failure modes are drawn from that mechanism rather than asserted: noise louder than the profile survives and is heard first in the gaps, noise quieter than the profile means the surplus subtraction comes out of the voice and produces the watery artefact people blame on the plugin, and noise that changes shape is wrong in both directions at once. A diagnosis step comes before any processing, because the fix is not worth the work if the noise was steady all along: compare three noise-only gaps from the start, middle and end rather than one, measure the level in each rather than trusting ears that adapt within seconds, and separate machines that cycle in and out at definite moments from traffic and weather that drift continuously. A false alarm is named explicitly, that an unchanged noise floor under a quieter passage of speech is a level problem and not a noise problem. The stock-REAPER method is then to split the recording into zones where the noise holds roughly still and profile each zone separately, with split points placed inside pauses, the Cockos wiki cited for S splitting selected items at the edit cursor, and two equally valid ways to carry separate settings: item FX, which the REAPER user guide documents in its "Item Take FX" and "Using Item FX with Individual Takes" sections, or one track per noise condition. Crossfades of roughly 20 to 50 ms are recommended at the joins on the grounds that an abrupt change in the noise bed is more noticeable than the noise itself, and the article notes that different amounts of subtraction change the perceived brightness of the voice and not only the noise, so levels need checking across each join. The limits of splitting are stated plainly: zones only exist if the noise holds still, and continuous drift gives forty joins to hide for very little gain, for which two honest options are offered instead, profiling the quietest stretch and accepting even lower-level noise because even noise is less distracting than noise that comes and goes, or a denoiser that needs no profile, with Jean-Marc Valin's RNNoise cited as a recurrent network that estimates noise from the signal over time rather than from a captured sample, and its bluntness disclosed. The honest-limits section states that noise changing inside a sentence cannot be split out because there is no gap to split at or profile from, that one-off events are edits rather than subtraction, that noise as loud as the voice cannot be separated from it, that room echo is not noise, and that a hiss identical on every recording ever made is probably the microphone or interface rather than the room. Prevention is treated as the only cheap fix and framed as a scheduling decision. REAPER is reported as having no documented way to rebuild a noise profile automatically during playback, with a long-running Cockos forum thread cited as evidence that the question recurs and that experienced users there know of no method, explicitly as context for the problem rather than as technical authority. No listening tests are claimed and no exact user-guide menu wording is quoted for the item FX step, because the current guide could not be retrieved in full. All sources were checked on October 10, 2026.
Home Vocal Recording and MixingHow to set up a headphone mix for recording vocals in REAPER
A take can come out flat, sharp, timid or over-sung purely because of what the singer was hearing, and the article's position is that the home-studio version of this problem is usually not a balance question at all but a routing one, because there are two separate paths the voice can take to the headphones and most people do not know which one they are on. It opens with a ten second test: record-arm the track, talk into the mic and move the track fader. If the voice changes level you are monitoring through REAPER and the faders are the cue mix controls; if nothing changes you are on the interface's direct monitoring, which Focusrite's own support documentation describes as taking the input and sending it "straight to the headphone and line outputs on the device" with "near-zero latency", so REAPER never sees it and cannot change it. The consequence the article is most concerned with is drawn from a REAPER forum thread in which a user, unable to hear enough of himself, had been raising preamp gain instead and was recording, in his words, "way too hot": a monitoring problem solved with the gain knob becomes a recording problem that cannot be undone. A four step build order follows, chosen because each step changes what the next should be: set the preamp for a clean recording and leave it, set the overall headphone volume using the voice alone with the backing muted, bring the backing up underneath only until the groove and key are clearly audible, and from then on change balance rather than volume. On the balance itself the article declines to give a ratio and cites Sound On Sound's position that the cue mix "should be one the performer wants to hear, not necessarily the ideal balance", together with its observation that headphone level "can make quite a difference, both to the quality of performance and to the singer's ability to pitch accurately". Where published guidance is genuinely actionable is in direction: Sweetwater's engineering guide advises that a consistently flat singer be given less of themselves, "try turning their vocal down a bit in the headphone mix", so they push harder, and that a consistently sharp singer be given more so they do not have to push. That is explicitly labelled practitioner guidance rather than a measured result, and flagged as useless when the pitch error is random rather than directional. A five row table maps what is heard in the take to the monitoring cause and a starting move, including the two cases that are not balance problems at all: timing drifting behind the beat, which is latency, and leaning in and out of the mic, which is a level that was set wrong. A real disagreement between two good sources is disclosed rather than smoothed over. Sound On Sound says a temporary reverb in the cue "usually helps inspire a more confident performance and can help with pitching"; Sweetwater says reverbs and delays "can make pitch issues harder to detect" and to start with little or no effects. Both are judged right about different risks, and the resolution offered is to use the smallest amount that makes the performance work and to check a doubtful take with the monitoring reverb bypassed. The routing section addresses the cost nobody mentions: if you monitor through REAPER, the faders you shove around for singing are the same faders you will mix with later. Three options are given in increasing order of effort, noting fader positions, using the routing matrix per-send volume described in the REAPER user guide's headphone monitoring section, which calls it useful "when a musician who wants to particularly hear only some tracks when recording", or building a monitor bus with Master Send unticked and a hardware output send. The reassurance that the fader is safe to move is sourced structurally rather than asserted: recording the post-fader signal is a separate mode the user guide documents as selecting Record: output from the record arm context menu, with output mode options of post fader, post-fx/pre-fader or pre fx, so a user who has not chosen it is not affecting the file. The user guide's Record disable (input monitoring only) option is offered for setting the balance up without recording anything. The one-ear-off section treats the technique as normal, citing Sweetwater on sliding the earpiece back while keeping it against the head and trying both sides, and Sound On Sound on the cost, that you need a way to mute the unused phone "otherwise you're inviting spill problems". On level, no decibel figure is claimed for any reader's headphones; instead NIOSH's recommended exposure limit of 85 dBA over an eight hour shift and the 3 dBA halving rule are cited as the scale, and two rules are derived that need no meter: set the level once with your own voice, and fix clarity with balance rather than volume. The limits section states that a good cue mix will not fix latency, will not improve a signal the preamp already got wrong, will not make anyone sing in tune, cannot rescue an existing take, and is not a mix. The honest expected outcome is given as fewer retries rather than better single takes. Forum threads are cited as evidence the problem is common and explicitly not as technical authority. All sources were checked and every quotation verified against the live page on October 9, 2026.
Practical MixingWhy your mix sounds bad in the car (and what the car is actually doing)
A mix that passes on headphones and monitors can fall apart at 60 mph, and the usual response, going back to the project and cutting bass, is often the wrong one. The article's position is that three or four separate things change the sound in a car and only one of them is the mix, so the car test is worth running only if it is set up to tell them apart. The first is the cabin itself. A car interior is small enough that its lowest acoustic modes land on top of the bass range: one published simulation and measurement study of a coupe interior puts the first longitudinal and lateral modes at 64, 73, 108 and 115 Hz, and notes that below a Schroeder cut-off of about 300 Hz wave acoustics dominate with only a few widely separated modes, with mode frequencies varying by vehicle type. Two practical consequences are drawn from that rather than asserted: low-end behaviour depends on where the listener is sitting, because the seat is inside a standing-wave pattern, and it differs between vehicles, so there is no such thing as mixing for the car. The diagnostic offered is that a boom on one or two specific notes in one car is a cabin mode and should not be equalised for, while a heavy low end that also shows up on an unrelated system is the mix. The second is masking, and this is the part most car-test advice omits. Road noise in a modern vehicle typically occupies about 20 Hz to 1 kHz, with most tire and road interaction noise between 100 and 500 Hz, mostly structure-borne below about 300 Hz, plus tonal tire cavity resonances around 200 to 250 Hz, and low-frequency components usually dominant in cabin perception. That band is where kick, bass and the body and weight of a lead vocal live, so at speed the vocal is not lost but covered in the range where it carries weight, and the common reaction of brightening it is identified as wrong: masking at 200 Hz is not fixed by adding 3 kHz. The third is the head unit. Manufacturer owner's manuals are cited directly: Mazda's ALC automatically adjusts audio volume and sound quality according to vehicle speed, Bose AudioPilot continuously adjusts the music to compensate for background noise and vehicle speed, and Volvo's volume compensation increases volume in relation to the speed of the car. The article draws attention to Mazda's wording, sound quality rather than volume alone, as the reason a car is not a reference. The fourth is that the audio is usually a normalised stream rather than the exported file, with Spotify's own documentation cited for adjustment to -14 dB LUFS per ITU 1770 and selectable playback levels of -11, -14 and -19, and the level-dependence of perceived tonal balance deferred by internal link to the reference-track article rather than repeated. A six-step test protocol then controls for all four: play a local file rather than a stream, switch the car's automatic volume and noise compensation off and tone controls flat, listen parked with the engine off first to establish a baseline, drive the same passage at the same level so that whatever changes is attributable to masking, move position or swap seats to expose standing waves, and confirm on a third unrelated system before changing anything. A five-row symptom table maps what was heard to cause, test and a starting move, including the explicit entry that a single-note boom in one car calls for no action at all, and ReaEQ band types are cited from the Cockos documentation with the moves given as starting ranges rather than settings. The limits section states that this will not make a mix sound the same in every car, will not survive the car's own processing if the driver leaves it on, will not substitute for a room where the low end cannot be heard, cannot recover genuinely masked detail, and is not a loudness test because what reaches the car has been normalised. The honest expected outcome is given as a shorter to-do list: usually one real finding plus two or three symptoms that turn out to belong to the vehicle. The cabin mode frequencies are explicitly flagged as one measured vehicle and not presented as typical, no cabin gain slope is claimed because no authoritative general figure was retrieved, and no listening tests are claimed. All sources were checked on October 8, 2026.
Home Vocal Recording and MixingHow to punch in and fix one bad word in REAPER (without hearing the edit)
One wrong word in an otherwise good vocal take does not mean re-recording the whole performance, and the replacement take is rarely as good as the one already captured. The article's position is that the record mode is the easy part and the matching is the hard part: REAPER's punch recording is non-destructive by default, creating a new take rather than destroying the original, so the real risk is not losing audio but producing a repair the listener can hear. Five causes of an audible punch are separated out, and four of the five are setup and performance problems rather than software problems: level mismatch, distance and tone mismatch from the proximity effect, a changed room and noise floor, flat delivery from dropping in cold, and a hard seam with no crossfade. The REAPER setup is given against the v7.81 user guide: Record mode time selection auto punch, overlapping recordings left on split existing items and add takes, repeat and loop points off, and plain tracks rather than fixed item lanes. Pre-roll is covered as the fix for cold delivery, with REAPER's default of two measures and a note that slow material often wants four. Monitor Input and Monitor Input (tape auto style) are distinguished. The matching section covers rebuilding mic distance and angle, leaving interface gain alone, and matching the room. The seam section reports that REAPER's shipped Item Fade Defaults give recorded media items a fade but no crossfade while split items do get one, which explains why a punch can sound abrupt where an ordinary split edit does not, and suggests manual crossfades of roughly 5 to 20 ms. An honest-limits section states what a punch cannot fix: a join inside a held note, a performance problem across the whole take, a tone mismatch that was not prevented, and a changed noise floor.
Practical MixingHow to use a reference track in REAPER (and level-match it first)
Dragging a finished record onto the timeline next to an unfinished mix is one of the most useful things a home mixer can do and one of the easiest to do wrong, because the record is typically several decibels louder and at that point the two are not being compared at all. The article's position is that a reference track is worth nothing until it is level-matched, and that the matching step is arithmetic rather than taste. Two mechanisms are given for why an unmatched reference misleads, both sourced rather than asserted. The first is that hearing does not have the same frequency response at every level: ISO 226:2023, third edition, specifies the normal equal-loudness-level contours across 20 Hz to 12,500 Hz, and because those contours flatten as level rises, the same material played louder presents relatively more low and high frequency content, so it sounds fuller and more open before anyone has mixed anything. The pure-tone basis of that standard is disclosed, so it is offered as the direction of the effect rather than a curve to apply. The second is that the remembered version of the reference was probably loudness normalised, with Spotify's own documentation cited for normalisation to -14 LUFS and a recommended true peak below -1 dB, or below -2 dB for masters louder than -14 LUFS, which means the app version and the imported file are at two different levels. The method is seven steps and uses only stock REAPER. Pick a reference in the same style that you own, because a streaming app cannot be imported, and prefer two or three over one so the comparison is not against a single engineer's taste. Put it on its own track, outside any folder or bus carrying processing, and leave it muted so it never reaches a render. Bypass the master FX chain while A/B-ing, because otherwise your own master processing is being applied to the reference and quietly making it resemble your mix. Measure the reference by selecting the item and running the action that calculates the loudness of selected items source media via a dry run render, which reports figures without writing media to disk. Measure the mix with a dry run render of the master with loudness calculation on, with a time selection if a shorter span is wanted. Subtract, and lower the reference by the offset rather than raising the mix, which would only move toward clipping and change what is being judged. Then A/B in short bursts asking one question per pass. REAPER's version history is used as the source for the measurement features rather than a claim about the current interface: 7.80 added actions to calculate the stereo loudness of selected tracks via dry run render and 7.75 is the version whose changelog names the selected-items equivalent, with 7.82 confirmed as current on October 4, 2026. A caveat most reference guides omit is included, that integrated loudness averages everything inside the span measured, so a full song against a thirty second chorus loop is not a like-for-like reading and comparable sections should be used. A five-row table then orders what to compare once the levels match, from lead vocal balance down through low end, top end, width and density, with each row naming what a difference usually means and flagging top end and density as the ones most likely to be mastering rather than mixing. The limits section is blunt: a reference will not make a mix as loud as a master, because that loudness comes from mastering done after the mix; it will not transfer tone, and copying a frequency curve off a finished master usually makes things worse; it will not fix arrangement, a room or headphones; and it cannot be reverse-engineered, because the imported file has been through a mix, a mastering chain and often a format conversion. The honest expected outcome is stated as a shorter to-do list rather than a transformation, most often that the mix was closer than it seemed and the vocal needed another decibel. A five-step troubleshooting order closes the article, starting with re-measuring the match and confirming the master is really bypassed before any plugin is opened. The LUFS figures in the worked example are labelled as illustrative arithmetic, and no listening tests are claimed. All sources were checked on October 6, 2026.
Home Vocal Recording and MixingWhy your voice sounds worse in the recording than it does in your head
Two separate things are happening when a home recordist hears their first playback and decides their voice is the problem, and only one of them can be fixed. While you sing or speak you hear yourself through the air and through the bones of your skull at once, and the article sources that split to a University of Tokyo feature by acoustics researcher Miki Yonemura, which describes normal self-hearing as a mixture of both routes and a recording as containing only the air-conducted component, meaning the recording is the version other people have always heard. The direction of the difference is sourced separately, to Imperial College London's report of research by T. Tchumatchenko and T. Reichenbach published in Nature Communications, which found bone conduction more effective at transmitting lower frequencies, with Dr Tobias Reichenbach stating the consequence plainly: we perceive our own voice as deeper than it is. The practical conclusion drawn from that is a warning rather than a technique, because if the missing weight arrived by a path the microphone never saw, trying to EQ it back in below roughly 200 Hz produces mud instead. The article's unique contribution is two tests that separate the permanent perception gap from a genuine capture fault before any money is spent: play twenty seconds of the raw take to someone who knows your voice and ask only whether it sounds like you, not whether it sounds good, and compare the raw take against a released track through the same system at roughly matched loudness. Both tests are specified on the raw take, because testing a processed one tests mix decisions instead. Only if a test fails does the article move to fixes, in order of effort: distance and angle, with DPA Microphones cited for proximity effect existing on gradient microphones and not on omnidirectional ones and appearing under about a metre, then the room, then level and a basic stock chain. A four-row table maps what the reader notices to a cause and a next step, and the limits section is explicit that no fix makes playback match the internal voice, that processing does not change the instrument, and that the buying question is answered elsewhere. NPR's February 2025 segment with voice researcher Rebecca Kleinberger is cited only as evidence that the question is widely asked, not for the mechanism. All sources were checked on October 5, 2026 and no testing is claimed.
Tools and RecommendationsThe best free LUFS meters for voice (and when REAPER's own is enough)
The honest answer to "which free LUFS meter should I install" starts by questioning the premise, because for the most common version of the job the answer is none. REAPER's render dialog can calculate loudness as part of a render and can do it as a dry run that reports the figures without writing media to disk, so checking what a finished episode measures needs no plugin at all. That is sourced from the Cockos changelog rather than asserted: 7.80 added actions to calculate the stereo loudness of selected tracks via dry run render, 7.75 added the equivalent for the source media of selected items, and 7.81, released September 28 2026, added the option to render with no loudness calculation at all. REAPER also ships its own loudness meter JSFX, documented in section 6.18 of the User Guide. The article sets its criteria before naming any product, which is the house rule for comparison pieces: genuinely free with no time limit, measuring to ITU-R BS.1770 and EBU R 128 rather than a house approximation, running as a plugin in REAPER on Windows or macOS, showing integrated plus at least short-term, and leaving the audio alone. A standards section explains why reading one number when the meter shows three is where most LUFS confusion comes from: momentary over a sliding 400 ms window, short-term over 3 seconds, and integrated across the whole programme with an absolute gate at -70 LUFS and a relative gate 10 LU below the running value, with no gating on momentary or short-term, which is why a conversation full of pauses still reports a sensible integrated figure. Loudness range is described as a statistical measure the standard sets no maximum for. Three free meters are then compared on those criteria. Youlean Loudness Meter 2's free version carries the LU, integrated, true peak, loudness range, peak to loudness and dynamic range readouts as VST2, VST3, AU and AAX, and the article is specific about what the $39 Pro tier holds back, naming the streaming presets as the one that actually stings because those are what put a target line on the meter. TBProAudio dpMeter 5.3.9 is the pick for spoken word and the reason is a feature no generic roundup highlights: alongside RMS and full EBU R 128 it offers a dialogue-gated mode reporting dialogue-gated integrated loudness, dialogue proportion and dialogue-gated loudness range, which describes the speech in a programme rather than the average of speech and music, and it is free with no paid tier, nag screen or registration. MeldaProduction's MLoudnessAnalyzer is the third option, EBU R 128 and ITU-R BS.1770-3 compliant with the EBU+9, EBU+18 and EBU+27 scales, with the practical caveat stated plainly that it arrives inside the free MFreeFXBundle rather than as a single download. A pick-in-one-minute table maps five situations to a choice. The limits section is blunt: every meter here is analysis only and Youlean's own documentation says the plugin cannot alter audio loudness, so reading -19 LUFS does not move you to -16; a file can measure exactly on target and still be noisy or harsh; integrated loudness is one figure for a whole file so a quiet guest and a loud host average out to something that looks right and sounds uneven; hitting a target does not stop a platform renormalising playback; and ACX's requirements are not written in LUFS, so a podcast-normalised file can still fail an audiobook submission. A troubleshooting section treats two meters disagreeing as a question about what was measured rather than about the meters, listing different measured spans, comparing integrated against short-term, an un-reset integrated reading, plugin placement before or after a limiter, and mono versus stereo. Delivery targets are attributed rather than invented, citing Apple's recommendation of around -16 dB LKFS with about 1 dB of tolerance and true peak not exceeding -1 dB FS per ITU-R BS.1770-5. One limitation is disclosed in the sources rather than papered over: the body text of User Guide section 6.18 could not be retrieved on the day of writing, so the article describes REAPER's documented render-time loudness calculation instead of listing the stock meter's on-screen readouts. Vendor pages for all three meters, the Cockos download page and changelog, the EBU R 128 measurement definitions and Apple's podcast audio requirements were all checked on October 4, 2026, and the article states that StudioRescue has no commercial relationship with any vendor named and carries no affiliate links.
REAPER Audio WorkflowsHow to batch convert and normalize a folder of audio files in REAPER
A folder of files at the wrong sample rate, at wildly different levels, or in the wrong format is a job REAPER already has a tool for, and it is not the render dialog. The article opens File, Batch File/Item Converter and goes through it control by control against section 21.11 of the REAPER User Guide v7.81. Add takes files from disk or selected media items from the current project, and the list shows channel count, sample rate and bit depth so the odd files out are visible at a glance; Presets save an output setup for next time. Output can go to the source directory or a chosen folder, and the Overwrite original files option is flagged as the one setting in the window that can cost you the only copy of something. File naming uses wildcards, with the default $source-converted turning Vox 1.wav into Vox 1-converted.flac, and the right-click menu can add the converted files straight back into the project. Sample rate and channels default to the source, and the explode behaviour is spelled out: mono to stereo duplicates the material to both channels, a stereo file exploded becomes two mono files, and a six channel surround file becomes six. The guide's own caution that resample mode can affect pitch, length and tempo is carried over rather than softened. The normalize half is the Normalize/trim/fade button, which opens post-process render options covering peak and loudness-based normalization including true peak and LUFS targets, brickwall limiting to a dB ceiling, leading and trailing silence trimming, padding, and fade curves, with the adjacent checkbox toggling all of it on or off. Targets are given as starting points tied to the destination rather than as rules: around -16 LUFS integrated as a common stereo podcast convention rather than a universal specification, and a peak target for sound effects where headroom matters more than perceived loudness. Use FX applies a saved FX chain to every file, with an FX tail field in milliseconds so reverb and delay are not cut off, and the article warns that a chain tuned on one voice in one room will not suit a mixed folder. The limits section is explicit: the converter trims silence from the ends but cannot remove a cough from the middle, normalizing sets level but does not even out a performance that swings, clipping and noise survive conversion intact, lossy to lossy re-encodes, and REAPER is paid software at $60 discounted or $225 commercial with a 60 day full-function evaluation, with the free command line tools ffmpeg and SoX named for conversion-only jobs. A closing section addresses the specific reason many people are looking for this in 2026: Audacity 4.0 shipped without Macros, confirmed on the official Audacity forum in September 2026, and REAPER's converter is offered as a good answer for converting and normalizing a folder but explicitly not as a Macros equivalent, because Macros replay an arbitrary sequence of editing commands and the batch converter does not. Checked against the REAPER User Guide v7.81 section 21.11, the Cockos downloads and purchase pages, and two Audacity forum threads on October 3, 2026.
Podcast and Spoken-Word AudioHow to record a double-ender so the two tracks do not drift out of sync
A double-ender solves the quality problem of a remote interview and introduces a quieter one that people discover halfway through the edit: the clap at the top lines up perfectly and the laugh forty minutes in lands late. The article's first job is to separate two failures that get called the same thing. A nominal sample rate mismatch is a large constant error present from the first second with an obvious pitch shift, and it has its own fix. A clock accuracy difference is what drift actually is: both ends recorded at the rate they were set to, but the crystal oscillator in each machine does not run at exactly the nominal frequency, so the files start together and separate slowly. The mechanism is sourced rather than asserted. ASIO4ALL's documentation states that whenever two clocks are not identical devices will slowly drift apart, and that a 100 ppm difference means two recordings will diverge by 360 ms every hour. Dan Lavry's paper on clock accuracy supplies both the reason nobody hears drift as a tuning problem, since a shift of one cent, under 500 ppm, is below the audible limit, and the reason the goal is not zero, since holding two independently clocked tracks within one microsecond over three minutes would require total error inside 0.0056 ppm. A table converts clock difference into separation per minute and per hour, and the top of that table is not hypothetical: a peer-reviewed MultiMedia Modeling 2015 study measured oscillator error across sixteen consumer hand-held devices and found misalignment reaching about 60 ms per minute in the worst case, with drift that was not even constant because the devices needed roughly ten minutes to reach working temperature. That study's scope is stated plainly, because it tested phones, tablets, media players and camcorders rather than audio interfaces, so it describes what consumer hardware can do and not a prediction for any particular rig. The prevention half is six setup steps: agree one sample rate in writing and set it in File, Project Settings and in Options, Preferences, Audio, Device; record locally from one wired input per machine; have the host record the call itself as the reference the two good files get aligned against, which is the approach the free Double Ender utility is built on; mark the end of the session as well as the start, which is the single most useful item in the list because a head marker only shows that something went wrong while a head and a tail marker let you calculate the rate; record one continuous file rather than pausing; and run a sixty-second test first, because the only drift figure that applies to your gear is the one you measure. Measurement is then arithmetic rather than judgement: align the head markers, read the gap between the tail markers, divide by the elapsed time, and a worked example of 420 ms over 58 minutes comes out at about 121 ppm, or roughly 7 ms per minute. It deliberately declines to publish a tolerance threshold and gives the reasoning instead, that conversational timing is far more forgiving than lip sync, and it warns that because drift is not linear a single stretch leaves residual error concentrated in the first ten minutes. Blunt about the limits, because it will not make drift zero without a shared clock that a remote interview does not have, will not repair a recording already made, will not fix a nominal rate mismatch, will not improve how the guest sounds, and will make a file that lost samples worse rather than better if it is stretched. Checked against ASIO4ALL device-aggregation documentation, Guggenberger, Lux and Boszormenyi in MultiMedia Modeling 2015, a Lavry Engineering clock accuracy paper, the Double Ender utility page and two official Cockos guides on October 2, 2026.
Home Vocal Recording and MixingHow to record vocals quietly in an apartment (without sounding timid)
Living somewhere you cannot be loud is one of the most common reasons a home recording never gets made, and the advice people are usually given, soundproof the room, is the one thing a renter cannot do. The article starts by separating the two jobs, because they get confused constantly. Acoustical Solutions states that acoustic foam does not block sound because it is porous and lets sound pass through, and that blocking instead takes mass, layers mechanically disengaged from one another, and sealing that is airtight. Panels, foam, reflection filters and fabric booths therefore change how you sound and not how much escapes, so the goal is reset from being inaudible to being much quieter while still sounding full. The lever that does the work is distance. Your voice rises steeply at the capsule as you move in, with the free-field point-source ideal being about 6 dB per doubling of distance, less than that in a real room because reflections fill some of it back in, while the two things you do not want, the apartment's own traffic and heating and appliance noise and the microphone's self-noise, do not rise with you. The ratio improves, and needing less preamp gain is the symptom rather than the cause. Capture is then set as starting ranges: about 4 to 6 inches, slightly off axis, with a pop filter used as a physical spacer, and gain raised until a genuinely quiet delivery peaks around -12 to -6 dBFS, with the loudest intended line rather than the average one setting the knob. The performance half is the part forum threads skip, and it is what decides whether the take sounds alive or apologetic: energy is read from throat and body tension, consonant edge, speed and commitment rather than from sound pressure, so a shout is acted at roughly a third of its real volume, diction is deliberately protected because consonants are the first thing to go when people get quiet, and the instinct to lean away on the big lines is the thing that undoes the whole setup. Working this close has one predictable cost, and it is sourced rather than asserted. Audio-Technica defines proximity effect as the exaggeration of low-frequency sounds in a directional microphone placed very near the source and notes omnidirectional mics are free of it, and Shure adds that too much of it makes vocals boomy or muddy. Because backing away is the one thing unavailable here, it is fixed afterwards in REAPER: a ReaEQ high-pass starting around 80 to 100 Hz, a 200 to 400 Hz peaking cut of 2 to 4 dB if the take is still thick, and ReaComp at around 3:1 with only 3 to 4 dB of gain reduction on peaks, deliberately light because a quiet performance already has a narrow dynamic range and heavy compression at this distance drags up every breath and the room behind you. Blunt about the limits, because none of this soundproofs anything, none of it removes noise happening while you record, it will not turn a whisper into a belt since a conversational-volume line lacks the throat compression and grit of a real full-voice take, and it is a recording method rather than a defence against a noise complaint. Checked against Audio-Technica and Shure manufacturer documentation, Physics LibreTexts on intensity and distance, Acoustical Solutions on absorption versus blocking, the Cockos ReaPlugs page and the REAPER User Guide 7.81 on October 1, 2026.
Audio Repair and RestorationWhy your recording hisses: is it the mic or the interface?
A steady hiss under every take is not one problem but three, and they have different fixes. The article separates the microphone's own self-noise, the preamp being pushed hard to make a quiet microphone usable, and the room and the computer in it, then gives a five-minute test that identifies which one you have: three ten-second clips recorded at identical gain, one with nothing plugged into the input, one with the mic connected in silence, and one with your voice at normal distance. Comparing them tells you whether the interface is your ceiling, whether the mic or the room added the noise, or whether the floor is fine and the signal is simply sitting too close to it, with an honest caveat that an unplugged input does not see the same source impedance as a connected mic, so clip one indicates the preamp's floor rather than measuring it. The numbers are sourced rather than asserted. DPA describes equivalent noise level as measured to IEC 60268-4 by two detector methods, A-weighted RMS and ITU-weighted quasi-peak, with the quasi-peak figure typically 11 to 13 dB higher for the same microphone, which is why two spec sheets are not automatically comparable. Neumann's published bands put below 10 dB(A) at extremely low noise, 16 to 19 dB(A) at good enough for most purposes and 20 to 23 dB(A) at clearly audible. DPA also notes that dynamic microphone self-noise is seldom specified at all, because the preamp usually determines the result. On the preamp side, Sound Devices defines equivalent input noise as output noise minus gain, with a floor of -133 dBV set by the thermal noise of a 150 ohm source, and the article reconciles that against DPA's -120 to -130 dBu target by pointing out that the two scales differ by about 2.2 dB, so the best physics allows is roughly -131 dBu. There is a correction at the centre of it: preamp gain does not add hiss, and it cannot improve the voice's ratio against anything upstream of the input, but adequate analog gain does lift the signal clear of the converter's own floor, which is why level belongs at the preamp rather than at a fader, and why normalizing a finished file improves nothing at all. Measurement uses REAPER's loudness calculation via dry run render, reading the gap between the clips rather than absolute values, and it deliberately declines to invent a signal-to-noise threshold because none is published for home recording. The fixes are ordered by how much they help, starting with working closer to the microphone, the only free change that alters the ratio. Blunt about the limits, because noise reduction cannot give back the breaths and quiet consonants the hiss was covering, will not rescue a take raised from far too quiet, and does nothing about tomorrow's session. Checked against DPA, Neumann, Sound Devices and Lewitt manufacturer documentation, the Cockos ReaPlugs page and the REAPER User Guide 7.81 on September 30, 2026.
Home Vocal Recording and MixingHow to record and place rap ad-libs in REAPER (so they support the main vocal)
Ad-libs are the second layer of a rap vocal, and when they stop working the symptom is specific: the main vocal is no longer the loudest and clearest thing in the song, and every word gets slightly harder to follow. That is almost always placement and level rather than performance. The article starts by naming the two kinds, because they behave differently. Reinforcements land on the lead, so they have to be tight in time or they blur the word they were meant to thicken. Responses land in the gaps, so they can be looser, wider and more heavily effected without costing anything. Then the order of operations. Record ad-libs last, against a comped lead, with the mic at exactly the same distance and gain, and split them by type onto their own tracks rather than one ad-lib track, because doubles, responses, pitched lines and shouts all want different treatment. Bus them in a folder track so every decision below is made once instead of four times. Place gaps first, level second, panning third, because changing that order is what sends people reaching for EQ on a problem that was really a volume problem. Starting ranges rather than settings: the bus around 4 to 8 dB under the lead, reinforcements lower still at 8 to 12 dB, and a high pass placed higher than the lead's at roughly 120 to 200 Hz, with a mute test as the check that actually decides it. Width is done in matched pairs, the same response recorded twice and panned apart, because a single hard-panned take disappears for anyone on one earbud. The effects section is the part that is sourced rather than judged: ReaDelay for throws, with delay length in milliseconds or fractions of a note and a per-tap lowpass to dull the repeats, ReaVerbate on a send to push a response behind a lead that sits right at the front, and ReaPitch for the low voice under a line, which Cockos describes as raising or lowering the pitch of an original sound with a shift in semitones, a separate cents control and a formant shift. When one ad-lib collides with one lead word, the fix is that word and not the whole track: a volume envelope across the phrase, or ReaComp on the ad-lib bus fed from the lead through the auxiliary input Cockos documents for audio ducking. Blunt about the limits, because ad-libs will not fix a weak lead, will not fix a muddy or harsh mix, will not create energy the performance does not have, and will not survive being as loud as the lead. Checked against the Cockos effects summary guide, the ReaPlugs page and the REAPER User Guide 7.81 on September 29, 2026.
Home Vocal Recording and MixingWhy your vocal sounds thin and quiet after uploading to YouTube, Instagram or TikTok
Your vocal sat right in REAPER, and the uploaded version came back smaller, duller and further away. In most cases nothing was removed from the recording: the platform measured how loud the file was and turned it down, and quieter playback is heard as thinner because human hearing loses low end and extreme top faster than midrange as level drops, which is what the equal-loudness contours in ISO 226 describe. The article separates three things that all get called “it sounds worse after uploading” and gives each a different fix. First, prove what changed: YouTube will tell you, in Stats for nerds, as a Volume / Normalized line with a content loudness figure that is the platform’s own measurement of your file against its reference, and a Dry Run render in REAPER gives you the integrated LUFS and true peak of the file you sent. Compare the two at matched level, because comparing an upload at one volume with a project at another is how people end up fixing problems that were never there. Second, crushed is a different problem and it is yours: heavy limiting flattens a vocal, the platform turns the loudness back down and does not undo the limiting, so pushing above the reference is a one-way trade that only costs you. Third, peaks that only appear after encoding: lossy delivery can reconstruct a waveform above your highest sample values, which is what true-peak metering in dBTP catches, and EBU R 128 sets a maximum permitted level of -1 dBTP while Spotify recommends a true peak below -1 dBTP for lossy formats and below -2 dBTP for masters louder than its -14 LUFS reference. There is also a deliberate refusal in it: YouTube, Instagram and TikTok publish no loudness specification, the commonly quoted estimates for YouTube disagree with each other, and no invented target is given for any of them. Checked against Spotify’s loudness normalization page, EBU R 128, ISO 226:2023 and the REAPER User Guide 7.80 on September 28, 2026.
Podcast and Spoken-Word AudioHow to fix mic bleed between two podcast mics in REAPER
Two people, two microphones, one room, and every track has both voices on it. Mic bleed is not a filtering problem, because the thing leaking into your microphone is speech in the same frequency range as the speech you are keeping, which is why notching works on a leaked click track and fails completely here. The article starts with the part most guides skip: why two bleeding tracks sound worse together than either does alone. Your co-host's voice now exists twice in the session, once close on their mic and once roughly a millisecond per foot later on yours, and the two copies partially cancel, which is the hollow, tunnelled sound people call phasey. There is a three-step test for how bad it is and whether it is cancellation or just clutter, then the fix that always works: cut each track down to its own speaker so the clean track carries every word, with automatic crossfades on, ripple editing off, and about a quarter of a second of that person's own room tone left either side of each turn. REAPER's Dynamic Splitting, documented in the user guide under the Item menu as removing silent passages, does the bulk pass on a long episode. ReaGate automates the rest, with a starting-range table that pairs each control with what Cockos actually says it does: threshold, attack, hold, release, hysteresis and pre-open, plus the rule that you tune a gate on the worst passage and not the easiest one. Blunt about the limits, because overlapping speech is where every method stops, including the paid machine-learning tools sold for this exact job, and a polarity flip cannot cancel bleed that arrives at a different delay for every sound. Ends at the microphone, where Shure's 3-to-1 rule and cardioid rejection prevent the problem far more cheaply than any plugin, with the honest note that the 180 degree null is a cardioid figure and supercardioid mics reject most a little off the rear. Checked against the Cockos effects guide, the REAPER User Guide 7.80 and two Shure publications on September 27, 2026.
REAPER Audio WorkflowsHow to hear reverb while recording a dry vocal in REAPER
Almost nobody sings well to a completely bare headphone signal, and the two usual beginner responses are both bad: sing dry and flat, or print reverb onto the take and lose the ability to ever change it. In REAPER this is a routing job rather than a plugin job. REAPER writes to disk what arrives at the armed track's record input, so anything downstream of that point, meaning sends to other tracks and the master, is heard and never written. That gives a two minute setup that works on any interface, including a USB mic with no second pair of outputs: add a track called Monitor Verb, send the vocal track to it, put ReaVerbate on the bus with wet up and dry off, and raise the send until singing feels comfortable. The article also covers the version for recording other people, a Monitor Bus with Master Send unticked and its own hardware output send to the performer's headphones, which is the build Sound On Sound's REAPER column documents, and why unticking Master Send is the step people miss. There is a starting-point table for ReaVerbate on a monitoring bus, a twenty second test that proves the recorded item is dry before you trust the routing, and the honest trade-off: monitoring through REAPER adds latency and every plugin in the path adds more, while hardware direct monitoring has none but bypasses REAPER entirely so there is no reverb in it. Blunt about what it does not do, because the take is dry by design and will sound bare on playback, reverb in the phones can mask real room echo that is being recorded, and monitoring settings are far too much for the finished mix. Checked against Sound On Sound, the Cockos effects guide, the Cockos troubleshooting appendix and the current REAPER release on September 26, 2026.
Audio Repair and RestorationCorrupted WAV file? How to recover the audio before you give up
A WAV file that nothing will open is usually not damaged audio at all. It is a damaged header, the short block of numbers at the front of the file that says how long the recording is and what format it is in, and the samples are still sitting on the disk behind it. A recorder cannot know the length of a take until you press stop, so it writes a placeholder and goes back to correct it at the end. Lose power before that moment and you get a file that may hold 400 MB of perfectly good audio behind a header claiming it is zero seconds long. This article gives the order to work in: stop writing to the card, work only on a copy, and try playing the file back on the recorder that made it first, because some recorders save at intervals during recording and can restore an interrupted file themselves on playback. Then two documented free routes. FFmpeg's WAV reader has an ignore_length option written for exactly this case, which skips the broken data chunk size and reads to the end of the file, and it belongs before -i because FFmpeg applies options to the next file named. Audacity's File, Import, Raw Data is documented for audio with no header, incorrect headers or partial corruption, and the four values you have to supply are recognisable when you guess wrong: static for the wrong bit depth or byte order, a fast stutter for the wrong channel count, wrong speed for the wrong sample rate. A symptom table separates a header problem from real data loss, from a rate label problem, from a REAPER path problem. Blunt about the limits, because nothing can create audio that was never written, rebuild a physically lost middle, decode a proprietary container, or recover a card that has been recorded over, and a 0 KB file is a disk recovery question rather than an audio one. Checked against Microsoft, EBU Tech 3285, the FFmpeg documentation, the Audacity manual and a manufacturer operation manual on September 25, 2026.
REAPER Audio WorkflowsREAPER keeps crashing or freezing? How to find the cause
A REAPER that crashes over and over is not usually mysterious, it is repeatable, and anything repeatable can be narrowed down. This article starts with the part most troubleshooting threads skip, which is describing the crash accurately: a crash during the plug-in scan, a crash the instant REAPER launches, a crash only in one project, a crash when one plug-in window opens, a freeze during playback after crackle, and a crash only on render each point at a different suspect. It then uses the three startup modifier keys documented in the REAPER User Guide to rule those suspects out one at a time, holding Ctrl to open REAPER without scanning new or updated VST plug-ins, Shift to open it without loading the last project, and Ctrl+Shift to open it without loading any default project template. Whichever switch changes the behaviour decides which of the three remaining sections you need. For a plug-in it covers halving your third-party folder to find the file, and REAPER's plug-in run mode under Options, Preferences, Plug-ins, Compatibility, which can give an unstable plug-in its own process so a crash in it does not take the session down, with the honest caveat that the menu wording has changed between versions and you should read your own install rather than an old screenshot. For a project it uses the Open with FX offline recovery-mode checkbox that Sound On Sound documents for loading a session that crashes on every normal attempt. It also shows how to make the next crash informative, because the Windows installer option that saves REAPER crash dumps is off by default, and on macOS Apple's Console lists crash reports under Crash Reports with the .ips extension. Blunt about the limits, because isolation will not fix failing hardware, will not make an abandoned plug-in compatible, will not recover the take you lost, will not explain itself in plain English, and cannot do anything with a crash that happens once a fortnight with no pattern. Checked against Cockos, Apple and Sound On Sound on September 24, 2026.
Audio Repair and RestorationHow to remove click track and headphone bleed from a vocal take in REAPER
Headphone bleed is the click track leaking out of the headphones and back into the microphone, and the reason it resists ordinary noise cleanup is that it is not a steady noise floor at all. A click is a short, loud, repeating event with silence between hits, which is exactly the signal shape that noise-profile tools are worst at. This article splits the problem the way it actually splits in practice: bleed in the gaps between phrases is fully removable with editing alone, using a time selection, Alt+Shift+A to select items within it, and Remove selected area of items so nothing later shifts, or Dynamic split for a take with many short gaps. Bleed underneath the voice is the hard case, and the most reliable stock tool there is a narrow ReaEQ notch at the click's own fundamental, found with a boost sweep during a gap, with the honest ceiling being much quieter rather than gone. It covers when ReaGate is worth trying, which is only when there is a real level gap between the softest word you want to keep and the loudest bleed, and why ReaFIR Subtract is the wrong tool for the click itself but the right one for any steady hiss left underneath. It also explains why the polarity-flip trick that comes up in every forum thread usually disappoints, because what reaches the mic is a filtered, delayed, quieter version of the click rather than a copy of it. Ends with the prevention list, which is unusually cheap compared to the repair: closed-back headphones, a lower monitoring level, one cup off, a duller click sample, and a ten-second silent test pass before the take. Checked against the Cockos ReaPlugs page, the REAPER Cockos Effects Summary Guide and the REAPER Accessibility Wiki on September 23, 2026.
Podcast and Spoken-Word AudioHow to cut a long podcast interview down to length in REAPER
An interview that ran 75 minutes for a 40 minute slot is a structural problem, not a filler-word problem, and the order you work in decides whether it takes an hour or a day. This article starts with the setting that nothing else works without: ripple editing all tracks, the mode that makes every track close the same gap so the host and guest stay locked together instead of sliding apart. It also covers the part almost every older guide predates, because REAPER v7.62 added a separate setting for whether ripple-all moves your regions and markers too, reachable by right-clicking the toolbar ripple button, and v7.74 moved ripple settings into their own Options submenu. Then the method: fix the target runtime and measure how far over you are, map the episode with named markers and regions and read their lengths in the Region/Marker Manager so you can see that one tangent is twelve minutes, and cut in four passes from whole segments down through slow answers and dead air to breaths and fillers last. The actual cut is a time selection plus Remove contents of time selection (moving later items), with auto-crossfades on so the joins do not click. Honest about what it cannot do, because removing material cannot make a dull conversation interesting, cannot supply an answer the guest never gave, does nothing about levels or loudness, and will drag your music bed along with everything else if you place it before the spoken edit is locked. Checked against the Cockos changelog, the REAPER Quick Start Guide and a professional Sound On Sound podcast workflow on September 22, 2026.
Audio Repair and RestorationAudio recorded at the wrong sample rate in REAPER
A take that plays slow and deep, or fast and chipmunked, is not damaged audio. The samples are intact and only the header number that says how many of them to play per second is wrong, which is why the whole problem is reversible. This article starts where most results do not, by ruling out the three REAPER settings that imitate a sample rate mismatch: an item playback rate that is not 1.0, a transport play rate that has drifted, and a beats-with-auto-stretch timebase reacting to a tempo change. It then gives the ten-second test that separates a file problem from a project problem, which is simply playing the file outside REAPER, and a table of the rate pairs behind nearly every case with the exact playback rate that reverses each one. A 48 kHz take mislabelled as 44.1 kHz runs about 8 percent slow and lands 1.5 semitones flat, and a playback rate of 1.0884 with preserve pitch off puts it back, after which a glue makes it permanent. For anyone who would rather not convert the audio at all, Audacity's Import Raw Data is documented as handling files with incorrect headers and lets you apply the true rate to the same samples. Blunt about the limits, because correcting the rate restores speed and pitch and does nothing about clipping, dropouts, a truncated file, or two recorders drifting apart over a long take. Checked against Microsoft, AES5-2018, Cockos and the Audacity manual on September 21, 2026.
Home Vocal Recording and MixingWill a better microphone actually fix my vocals?
A microphone captures your voice, your room, your background noise and the consequences of your distance and gain all in one file, and swapping it only changes the first of those. This article turns that into four tests you can run on the mic you already own in about ten minutes: record silence at your normal gain and then again with the HVAC and fans off to find out whose noise it is, record the same phrase at half your usual distance to find out whether the room is the limit, put a duvet behind the mic to confirm reflections, and read the waveform for clipping or a take recorded far too quietly. It also does the part the buying guides skip, which is what the published specs actually promise. Neumann's self-noise bands put 20 to 23 dB-A in clearly audible territory, but the same guidance notes that even a very quiet room contributes more noise than 10 dB-A, so your room and not your mic sets the floor in almost every home setup, and above roughly 8 mV/Pa a condenser's sensitivity stops being the limiting factor at all. The number worth remembering is from DPA's distance factor table: moving from a cardioid to a hypercardioid buys about 16 percent more working distance, while halving your distance is free and does far more. Blunt about the exceptions, because a new mic genuinely is the answer for a faulty or unpowered mic, a quiet room where hiss survives, and a real polar-pattern job. Checked against Neumann, DPA and Cockos on September 20, 2026.
Practical MixingHow loud should vocals be in the mix? (and how to check it in REAPER)
There is no universal dB figure for vocal level and no standards body or manufacturer publishes one, so this article refuses to invent one and gives the decision procedure instead. Three checks settle it: a reference track in the same style, a listen at conversation volume where only the loudest element survives, and a mono check, because phones and laptop speakers play something close to mono and stereo width hides balance problems. The piece then does the part almost nobody does, which is turning an ear decision into a number you can reuse: route the mix to a vocal bus and an instrumental bus, read each one with REAPER's stock JSFX loudness meter or a fast dry run render report, and write down the gap rather than the values, because the gap survives a master fader move. It also separates the three problems people mistake for a balance problem, namely performance dynamics, word-level swings and frequency masking, each of which has a different fix. Blunt about the limits: a fader cannot change tone, clean up a noisy room, repair a clipped take or make a track competitively loud. Checked against Cockos, Spotify for Artists, ITU-R BS.1770 and ISO 226 on September 19, 2026.
REAPER Audio WorkflowsREAPER says media offline? How to get your audio back
OFFLINE on a REAPER item is two completely different problems wearing the same label, and almost every result online only covers one of them. Either the audio is fine and REAPER has parked it because of a default option in Preferences and Media that releases files whenever REAPER is not the active application, or the files genuinely moved and the project needs pointing at them again. The ten-second test is to click into the REAPER window and watch: if OFFLINE clears, it is the setting, and if it stays, it is a relink job. The article covers both, plus the trap that catches people moving sessions between machines, which is that anything recorded before the project was first saved went to REAPER's default recording path and not the project folder. It is also blunt about what relinking cannot do: it is bookkeeping, not recovery, and it will not bring back a deleted file or fix missing plugins. Ends with the two habits that prevent it, saving before you record and consolidating with Save Project As before you archive. Checked against Cockos' own pages, The REAPER Blog and Sound On Sound against REAPER 7.80 on September 18, 2026.
Podcast and Spoken-Word AudioPodcast export settings in REAPER: format, bitrate, mono or stereo
For a normal talking show the answer is MP3, mono, 44.1 kHz, constant bit rate, somewhere in the 64 to 128 kbps range, with ID3 metadata embedded and the artwork kept small. Almost nobody needs a stereo podcast file and almost nobody needs 320 kbps. The article puts the published numbers side by side: Apple's own requirements give 64 to 128 kbps for mono RSS audio and 128 to 256 for stereo, and will not accept a single-channel WAV or FLAC upload, while Simplecast asks for 128 kbps or less so your file is not re-encoded. It also corrects a figure repeated everywhere: the -19 LUFS mono target does not appear in Apple's document, and the reason it coexists with -16 LUFS is that BS.1770 sums loudness across channels, so a one-channel file reads about 3 LU lower than the same audio across two. Finishes with the part that actually saves time each week, saving the whole configuration as a REAPER render preset. Checked against Apple's, Spotify's and Simplecast's own pages and the REAPER 7.80 User Guide on September 17, 2026.
Podcast and Spoken-Word AudioHow to meet ACX audiobook specs in REAPER
ACX rejects audiobook files on three numbers: RMS between -23 and -18 dB, peaks no higher than -3 dB, and a noise floor no higher than -60 dB RMS, delivered as a 44.1 kHz MP3 at 192 kbps CBR or higher. REAPER 7 measures and hits all of it with stock tools and no extensions: a dry run render reads RMS and peak, a time selection over room tone reads the noise floor, normalizing to RMS sets the level, and ReaLimit caps the peaks. Fix the noise before the level, because normalizing raises the room tone by exactly the same amount, which is why most home-recorded audiobooks come back. The article also corrects three figures repeated all over the search results that do not appear on ACX's current requirements page: the half-second head spacing, the mono-only rule, and the -6 dB peak floor. ACX also states that its own checker does not test your noise floor, so a clean report is not a pass. Checked against ACX's own pages and the REAPER 7.80 User Guide on September 16, 2026.
Tools and RecommendationsThe best free noise reduction plugins for voice (and when stock REAPER is enough)
Try ReaFIR first. It ships with REAPER, it is free, and spectral subtraction is the right design for steady hiss, fans and air conditioning. Beyond that, three genuinely free options cover the cases ReaFIR does not: Bertom Denoiser Classic works in the time domain with zero latency and needs no noise-only section to profile, Viator Voice De-Noiser gives you one knob for a fast podcast or stream pass, and the open-source RNNoise plugin removes far more noise on speech at the cost of sounding processed. The article explains the three denoiser designs so you can match one to your noise, gives a four-step way to judge any free denoiser in two minutes, and is blunt about the limits: none of these removes room echo, overlapping voices, mouth clicks or clipping, and every one of them trades voice quality for quietness. Availability and versions were checked on the developers' own pages on September 15, 2026. No affiliate links.
Podcast and Spoken-Word AudioHow to remove mouth clicks and lip smacks in REAPER
Mouth clicks and lip smacks are very short, fairly quiet ticks made by saliva and lips separating, and a close mic hears every one of them. No stock REAPER plugin strips them out automatically, so the reliable method is to find each one and reduce it rather than delete it. Clicks sitting in a gap between words come down with a take volume envelope or a small split and fade. Clicks sitting on top of a word need a spectral edit, using the region gain plus the time and frequency fades so the repair stays invisible. REAPER 7.80, released September 13, 2026, also added an Edit > Repair Pops/Clicks dialog that hunts for clicks automatically, which is worth running as a fast first pass. Noise reduction, de-essers, gates and narrow EQ cuts all miss mouth clicks, and the article explains why. Prevention at the mic beats every fix.
Practical MixingHow to prepare and export your vocals for a mixing engineer in REAPER
Handing your vocals to a mixing engineer is simpler than the jargon makes it sound. Send them dry: bypass your EQ, compression, de-esser, and reverb so the mixer can build the sound from a clean start. Export as 24-bit WAV or AIFF at your session's sample rate, not MP3, and set the render bounds to the entire project so every file starts at the same point and lines up on import. Keep each part as one continuous file, name your tracks before you render, and include a rough reference mix plus a short note with tempo and key. In REAPER, render stems with your plugins bypassed, or use Consolidate/Export for the untouched recordings. When in doubt, ask the mixer first.
REAPER Audio WorkflowsHow to recover a lost or crashed REAPER project
A crashed or lost REAPER project is rarely as bad as it feels. REAPER writes each recording to disk as a media file while you record, so even after a crash or a deleted project file, your takes are almost always still in the recording folder. To get the project back, look in the project folder for the newest backup, a .rpp-bak file or a timestamped copy, open it, and Save As a new name. If there is no project file at all, start fresh and re-import the audio from REAPER's recording path, then rebuild. After that, turn on auto-save and backups so the next crash costs you minutes, not hours. Everything here uses stock REAPER.
Podcast and Spoken-Word AudioHow to fix a podcast recorded in mono or on one channel in REAPER
If your podcast only plays in one ear, the voice was recorded onto a single channel of a stereo file, almost always because a mono mic was captured on a stereo track. The recording is fine, it is just parked on one side. In REAPER you fix it by setting the item's Take channel mode to the side with the voice, or by exploding two voices onto separate mono tracks, then re-exporting. A true mono recording that plays evenly in both ears needs no fix. Prevent it next time by arming a mono input per mic. Everything here uses stock REAPER.
Home Vocal Recording and MixingHow close should you be to the mic when recording vocals at home
How close you sit to the mic is one of the strongest tone controls you have, and it works before any plugin. For most home vocals on a cardioid mic, start about 6 to 10 inches away, roughly a fist to a hand-span, with a pop filter a few inches in front. Closer than that and the proximity effect makes the voice boomy and thick; farther and it gets thinner and picks up more of the room. There is no single right number: it depends on your mic, your voice, and how loud you sing, so record the same line at a few distances, level-match, and pick the one that sounds best. Distance cannot fix the room, the noise, or a shaky performance, but it costs nothing and everything here uses stock REAPER.
Audio Repair and RestorationWhy two vocal tracks sound hollow or thin together in REAPER
If two vocal tracks each sound fine alone but get thinner, hollow, or weaker in the low end when you play them together, that is a phase problem, not something an EQ can fix. It happens when a voice is captured by two mics, or when you stack a real double or a copied take, and the two versions do not line up. Try the polarity (phase) button in the track control panel first, which either fixes it in one click or rules polarity out. If it does not, zoom in and slide one take until its waveform lines up with the other. Better still, avoid it at the record stage with one mic, the 3:1 rule when you use two, or a real second take instead of a copied one. Everything here uses stock REAPER, so it costs nothing.
REAPER Audio WorkflowsREAPER not detecting your microphone or interface (how to fix it)
Plugging in a mic or interface and getting no input in REAPER is almost never broken gear, it is a setup problem you can usually fix in a few minutes. Check four things in order. First, let your operating system allow apps to use the microphone: on Windows 11 turn on Let desktop apps access your microphone under Settings, Privacy and security, Microphone, and on macOS enable REAPER under System Settings, Privacy and Security, Microphone. Second, in Options, Preferences, Audio, Device pick the right audio system, ASIO or WASAPI on Windows or Core Audio on a Mac, and install the interface driver. Third, enable the input range in the same panel, for example first input 1 and last input 2. Fourth, record-arm the track and set its input to Mono, Input 1, then watch the meter move. It will not repair a dead interface or cable, fix silent playback, or improve how a recording sounds, but for the common no-input problem this order finds the cause fast.
Home Vocal Recording and MixingWhy your vocal sounds robotic or over-processed in REAPER
A robotic or over-processed vocal almost always means one plugin is doing too much: pitch correction snapping too hard, a compressor squashing the dynamics flat, noise reduction turned up until the voice sounds watery, or several effects stacked past the point where they help. Bypass the whole FX chain, then enable plugins one at a time to hear which one turns the voice robotic, squashed, or watery, and A/B every change at matched volume so louder does not fool you. In REAPER you can fix all of it with stock plugins: slow ReaTune's attack time or use manual mode for pitch, aim for roughly 3 to 6 dB of gain reduction in ReaComp, subtract less noise with ReaFir, and go light on de-essing. It will not tune a badly off-key take, but for the common over-cooked sound the answer is almost always less processing, not more.
Podcast and Spoken-Word AudioHow to record a voiceover in REAPER for YouTube
A YouTube voiceover is a solo narration recorded to picture, and it needs a slightly different setup from a sung vocal or a two-person podcast. In REAPER, record one mono track at 48 kHz to match video, aim your loudest words around -18 to -12 dBFS so nothing clips, and monitor on headphones. Clean it with a light stock chain: a high-pass in ReaEQ near 80 to 100 Hz, gentle ReaComp for a steady level, and a de-esser only if the take needs it. Check loudness with REAPER's render statistics and aim the finished voice near -14 LUFS with true peaks at or below -1 dBTP, which is roughly where YouTube normalizes playback. Then render a 48 kHz WAV to drop into your video editor. A quiet room and steady mic distance matter more than any plugin, and everything here uses stock tools.
Practical MixingHow to make vocals wider in REAPER with stock plugins
Real stereo width needs different sound in the left and right channels, so the widest and most reliable result is a vocal you actually recorded twice, or separate harmony parts, panned apart, with the lead kept centered. A single lead vocal is usually meant to sit in the middle, and that is a feature, not a limit. You can still add apparent width to one take in REAPER with stock plugins: a light detune with ReaPitch panned to the sides, a short 10 to 30 ms delay on one side with ReaDelay for a Haas-style spread, or stereo reverb and delay around a centered dry vocal. Each trick that adds the most width also costs the most in mono, where phones and small speakers live, so sum to mono with the master track's mono button and back off anything that turns thin or hollow. Keep the moves small, stock plugins only, nothing to buy.
Home Vocal Recording and MixingHow to stop mic bumps, desk thumps, and handling noise in REAPER
Thumps and low booms in a take are usually handling noise: a knock, a footstep, or desk vibration traveling up the stand and into the mic, where most of the energy lands low. The reliable fix is to stop it at the source, because a bump recorded under a word cannot be cleanly removed later. Put the mic in a shock mount, get the stand off the desk or decouple its base, and keep your hands off the desk during takes. Then clean up what is left in REAPER with stock plugins: high-pass the low rumble with ReaEQ starting around 80 to 120 Hz, and split out or gate any thumps that land in the gaps between phrases. A noise gate only helps in the silences, since it is open while you talk, so the one case that resists repair is a loud thump directly under a word, which is usually faster to punch in and re-record.
Audio Repair and RestorationHow to remove background music from a recording
You cannot erase music from a mixed voice recording with normal noise reduction, because music is not the steady background sound those tools are built to subtract. Pulling a voice back out from under music is a separation job, and the realistic tool is AI stem separation: run the file through a free separator such as Ultimate Vocal Remover, which uses Demucs and MDX-Net based models on your own computer, then keep the vocal stem and clean it up in REAPER. Expect a real improvement on clearly produced music and a rougher result when the voice and music overlap, with some watery artifacts and a faint musical ghost. Trying two or three models and keeping the best output helps more than any EQ afterward. If you can find the original separate tracks, use those instead, because they always beat separating a finished mix.
REAPER Audio WorkflowsHow to automate vocal volume in REAPER
When a few words jump out or drop back, the fix is to change the level of just those spots instead of leaning harder on a compressor. REAPER gives you four tools that overlap: the track fader for the whole track, the item volume handle for a flat change to one clip, a take volume envelope for shaping level inside a clip, and the track volume envelope for riding level across the song. Even out the loudest one or two words with the item volume handle first, then draw a track volume envelope for the phrases that still drift, keeping moves to a decibel or two. You can draw the envelope by hand or record your fader moves using automation modes like Touch. It is a level fix, not a tone fix, and it will not replace compression when the level jumps on every syllable. Stock tools only, nothing to buy.
Home Vocal Recording and MixingHow to record and layer harmony and background vocals in REAPER
A single lead vocal can sound bare, and harmony and background vocals are what make a chorus lift, but stacking four or five parts often just sounds cluttered. The fullest, most natural result comes from recording each harmony as its own part on its own track, then blending them: pan matched pairs out to the sides, keep the group a few dB below the lead, and high-pass and thin their low mids so they support the lead instead of crowding it. Group the backing vocals in a folder so one fader and one reverb glue the whole stack together. REAPER can also fake harmonies from a single take with the stock plugins ReaPitch and ReaVoice, which is fast and fine for sketches or thickening, but a part you actually sing almost always wins on anything exposed. It all uses plugins that ship with REAPER, so there is nothing to buy.
Home Vocal Recording and MixingHow to fix a vocal that was recorded too quietly in REAPER
Recording a vocal too quietly is a level problem, not a ruined take. As long as the voice sits above the noise, you can bring it up to a proper level in REAPER with stock tools. Raise the take with the item volume handle, the Normalize button (peak, RMS, or LUFS), or gentle compression with make-up gain, then handle any hiss that came up with it and add a limiter to catch stray peaks. The one thing you cannot undo is the ratio of voice to noise, which is set the moment you record, so if the noise is nearly as loud as the voice, re-record instead. Next time, aim for peaks around -18 to -12 dBFS going in.
Tools and RecommendationsThe best free de-esser plugins for vocals (and when stock REAPER is enough)
You probably do not need to download anything to de-ess a vocal. REAPER can do it with the stock ReaComp compressor and its detector input filter, or with ReaXcomp's top band, and for a large share of home vocals that is enough. Learn the stock method first, because the concept carries everywhere: find the sibilance, usually around 5 to 9 kHz, compress only that band, and use as little reduction as you can. Add a dedicated free plugin only when you have a reason. Techivation T-De-Esser 2 is the fastest one-knob fix, with frequency-range modes and a listen filter, though it needs a free account to download. TDR Nova is a free dynamic EQ that lets you de-ess one exact frequency with precise control. Airwindows DeEss is a light, open-source, no-account option that catches sibilance by tracking slew rate instead of pure frequency. Skip abandoned 32-bit-only plugins like SPITFISH unless you already run a bridge. A de-esser only tames harsh ess and sh sounds; it will not fix a muddy, thin, clipped, or badly recorded vocal.
Podcast and Spoken-Word AudioHow to match loudness between your podcast host and guest in REAPER
A host and guest sitting at different volumes is the most common reason a two-person podcast sounds amateur, and in REAPER you can fix it with stock tools. The trick is to match perceived loudness, not peak level, so start by putting each speaker on their own mono track. Add the stock JS Loudness Meter to each track, play a real stretch of speech, and read the integrated LUFS value for each voice. Bring both to the same number, around -18 LUFS is a good working level, either by nudging each track's gain or by normalizing each speaker's items to LUFS one speaker at a time. Add a gentle ReaComp so the match holds through loud and quiet moments, then set the finished episode near -16 LUFS with true peaks under -1 dBTP. This will not separate two voices captured on one mic, repair a distorted guest, or make two different mics sound alike, but it removes the single biggest thing that makes casual episodes tiring to listen to.
REAPER Audio WorkflowsHow to save and reuse FX chains and track templates in REAPER
If you add the same plugins and dial in the same settings on every new vocal, REAPER can do that work for you. It has two features people mix up: an FX chain saves the plugins and their exact settings from one track as a .RfxChain file, while a track template saves whole tracks, including their FX, routing, and sends, as a .RTrackTemplate file. Use an FX chain to reuse a plugin stack on a single track; use a track template to reuse one or more entire tracks with their wiring. Save from the FX window or the Track menu, and both files land in REAPER's resource folder (Options, Show REAPER resource path), where you can organize them or copy them to another computer. Two things trip beginners up: a track template only keeps sends between the tracks you actually selected, so include the bus, and a chain only stores which plugins to load, not the plugins themselves, so build from stock plugins to stay portable. Saving your setup speeds up every session, but the recalled settings are a starting point, not a finished mix, and no template fixes a bad recording.
Audio Repair and RestorationHow to remove a cough or one-off noise in REAPER
A single unwanted sound, a cough, a chair squeak, a desk thump, or a page turn, can ruin an otherwise good take, and it does not need a noise-reduction plugin. The fix in REAPER depends on one thing: where the noise sits. If it lands in a gap between words, zoom in, split the item on both sides with S, and either mute that slice or patch the hole with a copy of nearby room tone so the silence does not stand out, letting REAPER's automatic crossfades hide the seam. If it lands on top of the voice, you can only reduce it, not erase it: a short volume-envelope dip for a quick thump, a narrow ReaEQ notch for a ringing squeak, or REAPER's basic spectral edit for a broadband burst. A subtractive denoiser like ReaFIR is built for steady background noise and will not remove a one-off. This will not cleanly lift a loud cough off a sustained note, REAPER has no dedicated spectral-repair engine, and when the noise cannot be separated the honest fix is to re-record that one line.
Practical MixingHow to use parallel compression on vocals in REAPER
Parallel compression thickens a vocal by blending a heavily compressed copy under the untouched original, so you add weight and consistency without flattening the natural dynamics. In REAPER you can do it two ways with stock plugins. The fastest is a single ReaComp: crush it with a low threshold, high ratio, and fast attack, then raise the Dry slider so the compressed and dry signals mix together. For more control, send the vocal to a second track, compress only that copy, and ride its fader up under the main vocal so the voice gains density and stops disappearing in quiet phrases. Start with heavy compression, blend the copy quietly underneath, and stop as soon as the vocal sounds fuller rather than squashed. This will not fix pitch, timing, or tone, it raises breaths and room noise along with the good detail, and it is not a loudness maximizer.
Home Vocal Recording and MixingHow to record vocals in an untreated room
Home vocals sound roomy because the mic hears the voice and the reflected room at the same time, and in an untreated bedroom you fix most of that for free at the source. Clap once where you plan to sing: a ringing, fluttery tail means reflections off bare parallel walls that will end up in the take. Stand off the dead center of the room and out of corners, then use the mic's own directionality by pointing a cardioid's back at your most reflective wall. Get closer than usual, about 4 to 8 inches with a pop filter, so the direct voice is louder than the room, and treat the small proximity boom as an easy tone fix later. The highest-value cheap treatment is soft mass directly behind the mic: a hung moving blanket, a wardrobe full of clothes, or a mattress on its end, aimed at the bare walls that caused the flutter. Record a mono track at -12 to -6 dBFS with a light high-pass around 70 to 90 Hz. This will not remove room sound already baked into an old take, turn a bedroom into a studio, or fix a vocal that just sits too quiet in the mix.
REAPER Audio WorkflowsHow to install VST plugins in REAPER (and fix plugins not showing up)
You installed a VST but it is not in REAPER's FX list, and almost always the plugin is fine: it just landed in a folder REAPER is not scanning. Installing a plugin is really two steps, run its installer, then make sure REAPER looks in the folder it installed to. VST3 files have a fixed home (Common Files\VST3 on Windows, Library/Audio/Plug-Ins/VST3 on Mac) and usually appear on their own, while VST2 (a .dll or .vst) has no fixed location, so it is the format that most often goes missing. The fix is to open Options · Preferences · Plug-ins · VST, add the folder where the plugin installed to the VST plug-in paths, and run Clear cache and re-scan. If it still will not show, check 32-bit vs 64-bit, make sure a VST3 sits in the Common Files\VST3 folder, and on macOS check the system did not block it. This will not conjure a plugin that was never installed, authorise a paid demo, or load a macOS-only AU on Windows. REAPER's own stock Rea-series plugins never need any of this.
Podcast and Spoken-Word AudioHow to edit out filler words and breaths in REAPER
Raw spoken takes are full of ums, restarts, gaps, and audible breaths, and REAPER has no stock filler-word detector, so the ums and false starts are a manual cut: place the edit cursor on each side of the word, press S to split, delete the middle, and let Ripple editing close the gap. Breaths are different. You reduce the loud ones rather than delete every one, because speech with every breath removed sounds robotic. Strip long dead air in bulk with Dynamic split or Auto trim/split using generous 50 to 150 ms pads so word starts are not clipped, and tame loud breaths by hand with a small volume dip or a gentle ReaGate with the dry signal blended back in. Work in order: cut filler, strip dead air, tame breaths, then listen across every edit for rhythm and clicks. This will not remove steady background noise, set your loudness, or fix plosive pops, and over-editing leaves a voice sounding breathless, so keep the natural micro-pauses. All with REAPER stock tools.
Home Vocal Recording and MixingWhy your vocal sounds distant in REAPER (and how to make it sound closer)
A vocal that sounds distant or far away almost always means the mic captured too much room and not enough direct voice, and then the vocal was left too quiet or missing presence. The biggest fix is at the source: record closer, about 6 to 12 inches, in a deader spot, which raises the direct-to-reverberant ratio that makes a voice feel close. In the mix in REAPER, bring the vocal level up and keep it dry, because a long reverb tail is exactly what makes a vocal sound far away, then add a small presence lift around 3 to 6 kHz in ReaEQ and use gentle ReaComp to pull the quiet detail forward. Name the problem first: an audible tail after words is reverb, not distance, and a weak, airless tone is thinness, not distance, so each has its own fix. EQ and compression will not remove baked-in room reverb or add detail a distant mic never captured, so a take drowning in room is a re-record. All with REAPER stock plugins.
REAPER Audio WorkflowsHow to export audio from REAPER (render settings for WAV and MP3)
Saving a REAPER project and exporting an audio file are two different things, which is where a lot of beginners get stuck. To turn a mix into a playable file, open File, Render (Ctrl+Alt+R on Windows, Cmd+Option+R on Mac), leave Source on Master mix, set the bounds and the folder, and pick a format. Render WAV at 24-bit for a master, an archive, or anything an engineer will work on again, and MP3 at 320 kbps for emailing or uploading; current REAPER makes MP3s with no extra download. The Secondary output format box lets you write both in one pass. Leave Normalize off for rough exports; for streaming, keep peaks under -1 dBTP and let the platform set loudness, because Spotify and the others turn loud masters down rather than up. Rendering will not fix clipping, an unbalanced mix, or a quiet level, so mix first, then always play the exported file back before you send it.
Home Vocal Recording and MixingWhy Your Vocal Sounds Boxy in REAPER (and How to Fix It)
Boxy vocals sound hollow and closed-in, like the singer is talking into a cardboard box, and the cause is almost always a resonant buildup in the low mids, usually somewhere in 300 to 600 Hz. That range also holds the natural body of the voice, so the fix is precision, not a big cut. Find the worst frequency with a narrow ReaEQ boost sweep, then turn that band into a gentle cut of a few dB. Add a high-pass around 80 to 100 Hz only if there is rumble underneath. Boxy is not the same as muddy (lower and heavier, 150 to 400 Hz) or nasal (higher and pinched, around 1 kHz), so name it before you reach for EQ. EQ will not remove room reflections baked into the take or fully fix a mid-heavy mic, and if the vocal is boxy only in the full mix the real fix is cutting the same range on the instruments underneath it. All with REAPER stock plugins.
Audio Repair and RestorationHow to remove wind noise and low rumble in REAPER
Wind and low rumble are two different problems and they need different fixes. Steady rumble under the voice, from traffic, HVAC, or a bumped desk, filters out cleanly with a high-pass in ReaEQ starting around 80 to 120 Hz. Steady low wind hiss can be reduced with ReaFIR in Subtract mode, building a noise profile from a gap with no speech, but only for constant noise. The hard part is gusts: the roaring, thumping air that hits the capsule directly is broadband and can even overload the mic, and no stock plugin fully removes it. Test a speech gap first to tell rumble from gusts, high-pass the rumble, subtract the steady wind, and edit or automate a low cut over the gusts. Filtering will not restore a clipped gust or remove buffeting that spreads into the voice, and for heavy wind a windscreen at the source beats any repair.
Home Vocal Recording and MixingHow to double track vocals in REAPER (stock plugins)
Double tracking means performing the same vocal part more than once and playing the takes together, which is one of the quickest ways to make a chorus sound wider and fuller. The best double is a second real take: record the part again on a new track, pan the takes apart (lead center with the double lower and to one side, or both takes hard left and right), and set the double a few dB below the lead so it supports rather than competes. Send the lead and doubles to one bus so you can mix the stack as a single thing. When you cannot re-sing, ReaTune on automatic correction can build a fake double from a single take, and a short ReaDelay copy in the 15 to 35 ms range adds width but is the most fragile in mono. Always sum the mix to mono and confirm the vocal does not thin out or cancel. Doubling will not fix tuning, timing, or tone, and it is often too much on a quiet verse.
Podcast and Spoken-Word AudioHow to add intro music and duck it under your voice in REAPER
An intro-music bed sounds great until you start talking and the music buries your voice. The fix is ducking: the music drops under your voice and lifts back up in the gaps. REAPER gives you two stock ways to do it. For a fixed intro and outro, draw a volume automation dip on the music track, which gives you exact control over the fade and the dip. When you talk over a bed for long stretches, set up automatic ducking with ReaComp: make the music track 4 channels, send your voice to channels 3/4, and set ReaComp's detector input to that sidechain so the music ducks itself whenever you speak. Start with the music about 6 to 12 dB under your voice, a ratio around 4:1, and a slower release so it eases back smoothly. Ducking will not fix a busy or vocal-heavy music bed, set your final delivery loudness, or clean up a noisy voice.
REAPER Audio WorkflowsREAPER CPU spikes and stuttering? How to lower CPU usage
Stuttering, crackling playback while REAPER's CPU meter spikes almost always means the real-time audio load is too high for your buffer size, usually from too many or too heavy plugins, not a dying computer. Open the Performance Meter with View, Performance Meter to see your real load and which tracks are heaviest, because the number that causes dropouts is REAPER's real-time thread, not your total system CPU. Then work least-destructive first: raise the audio buffer in Options, Preferences, Audio, Device, turn on anticipative FX processing under Audio, Buffering, share one reverb on a send bus instead of copying it per track, and freeze or apply FX on the heaviest tracks. Lower a very high sample rate if you are running one. It will not remove crackle baked into a recording, bring back total silence, or turn a slow computer into a fast one.
Home Vocal Recording and MixingHow to record vocals with a USB mic and no interface
A USB microphone is its own audio device, so you do not need an audio interface to record vocals in REAPER. Plug the mic in, open Options, Settings, Audio, Device and select it as the input, using the WASAPI audio system on Windows or Core Audio on Mac. Add a track, set it to a mono input, record-arm it, and set your level so peaks land around -12 to -6 dBFS using the mic's own gain or the operating system, since there is no interface preamp. Monitor through the headphone jack on the mic itself to avoid a latency echo, and turn REAPER's software monitoring off when you do. On Windows a USB mic and a separate interface cannot run at the same time. It will not improve how the mic sounds or remove latency if the mic has no headphone jack.
Practical MixingHow to use a vocal bus in REAPER
REAPER does not give you a separate bus track, which trips up people coming from other DAWs. In REAPER any track can be a bus. To control several vocal tracks with one fader and one set of plugins, put them in a folder track, which sums them into a submix bus. To share one reverb or delay across those vocals, add a post-fader send to a separate effects track, which acts as an aux bus. Do the per-voice fixes on the individual tracks first, and keep the bus for moves that belong to the whole stack. It will not repair a bad take or fix a wrong balance.
Audio Repair and RestorationHow to fix audio that is out of sync with video in REAPER
Out-of-sync audio has only two causes, and they need different fixes. A constant offset means the audio is early or late by the same amount for the whole clip, so load the video into REAPER, line up a sharp sound like a clap with the picture, and slide the audio into place. Drift means the audio slowly slides further out as it plays, which almost always comes from variable frame rate video off a phone or a screen recorder, and the reliable fix is to re-encode the footage to a constant frame rate first with a free tool such as HandBrake, then re-sync. Match your project frame rate and sample rate to the video. It will not fix recording latency, dropouts, or the sound of the take, only its position in time.
Home Vocal Recording and MixingWhy your budget condenser mic sounds bad (and how to fix it in REAPER)
A cheap condenser that sounds harsh, thin, boomy, or full of room is almost never a broken mic. Budget condensers are sensitive and bright by design, so they expose an untreated room and a rushed setup more than an expensive mic would. Before you blame the capsule, check three things: gain set so the loudest words peak around -12 to -6 dBFS, distance of about 6 to 12 inches slightly off to the side, and something soft killing the room reflections. Then clean up the tone in REAPER with a high-pass and a small EQ cut using stock plugins. A new mic fixes the capsule, not the room or the setup, so it will not rescue a bad room, undo clipping, or fix proximity boom.
Podcast and Spoken-Word AudioHow to record a podcast in REAPER, start to finish
The part that trips people up is not any single button, it is the order of operations. A whole episode in REAPER is six steps: set up one track per voice pointed at your interface, record with a few seconds of room tone, cut mistakes and dead air with ripple editing so tracks stay in sync, clean up with a high-pass, gentle ReaFIR noise reduction, and light ReaComp, add music if you want it, then level to about -16 LUFS with true peaks under -1 dBTP and render to MP3. It is all built in, and it will not rescue a bad room, undo call compression, or make a thin USB mic sound like a studio, so record clean first.
REAPER Audio WorkflowsREAPER has no sound? How to fix it when playback is silent
Opening a project to dead silence in REAPER is almost never broken gear. Split the problem in two. If the Audio system box shows Dummy Audio or no device, pick a real one in Options, Preferences, Audio, Device, using ASIO for an interface or WASAPI otherwise. If a real device is selected and the meters move but you still hear nothing, it is a mute or routing problem, so unmute, route the track to the Master through channels 1/2, and check your system output. Usually a five-minute settings fix, and it will not repair a silent recording, an empty take, or a disconnected interface.
Home Vocal Recording and MixingHow to gain stage vocals in REAPER (set the right recording level)
Gain staging a vocal is one decision: set the recording level so the loudest part peaks with room to spare and never hits 0 dBFS, where clipping happens. Record in 24-bit, then set the gain on your audio interface, not REAPER's fader, so the loudest sung part peaks around -12 to -6 dBFS with the average near -18 dBFS. That leaves headroom so you never clip, and 24-bit is quiet enough that you never need to record hot. It will not repair a take that already clipped or change the tone of the voice.
Tools and RecommendationsThe best free reverb plugins for vocals (and when stock REAPER is enough)
For most home vocals, the reverb you already have is enough: REAPER's stock ReaVerbate on a send, set to a plate or small room, covers the everyday job. Add a free third-party reverb only when you want a specific character, and the ones worth downloading are OrilRiver for natural rooms and halls, TAL-Reverb-4 for classic plate sheen, Dragonfly Reverb for a full hall/room/plate bundle, and Valhalla Supermassive for long ambient washes. All four are genuinely free with no time limit and run in REAPER on Windows and Mac. No reverb will fix a bad recording or remove reverb already baked into a take.
Audio Repair and RestorationHow to fix crackle and clicks in a recording in REAPER
Before you fix anything, find out which kind of crackle you have. If it lands in the same spot on every playback, it is baked into the file: remove a few isolated clicks by splitting the item around each one so REAPER's tiny edge fades smooth the seam, or dip the volume over the spike. REAPER has no stock declicker, so send dense, continuous crackle to a free tool like Audacity's Click Removal. If the crackle only appears while you record or play back and moves around, it is not in the file at all, it is your buffer or CPU, and you fix it in settings. Some damage, like missing samples from a dropout or heavy clipping, cannot be fully restored.
Home Vocal Recording and MixingHow to remove plosives (p-pops) from vocals in REAPER
That punch of low-frequency boom on a p or b sound is a plosive, and it is easy to fix in REAPER without buying anything. For a few clear pops, isolate each one by splitting the item around it, then add ReaEQ as an item FX with band 1 set to high-pass around 120 to 150 Hz, or simply dip the volume over the pop. For lots of small pops, a gentle track-wide high-pass around 80 to 120 Hz works, but push it too far and the voice goes thin. Stock plugins only, and no filter will rescue a pop so strong that it already distorted the recording.
Podcast and Spoken-Word AudioHow to record two people in one REAPER session (two mics, separate tracks)
Recording a co-host or guest in the same room? The hard part is not REAPER, it is getting two mics in as separate channels. Two separate USB mics usually will not work together, because the computer talks to one audio device at a time, so use an interface with two inputs. Then add two tracks, set them to Mono Input 1 and Mono Input 2, record-arm both, set each level to peak around -12 to -6 dBFS, and record. You get two independent takes to edit and mix, though separate tracks will not remove mic bleed or fix a bad room.
Practical MixingHow to add delay to vocals in REAPER (with stock ReaDelay)
Want an echo or a fatter lead vocal? REAPER already ships with ReaDelay, a capable multi-tap delay. Put it on its own send track with Dry pulled all the way down, set the length in milliseconds for a slapback around 80 to 160 ms or to a musical note value so the repeats land in time, keep feedback low, and filter the echoes so they sit under the voice. Stock plugins only, and delay is a creative effect, so it will not fix timing, pitch, tone, or room echo baked into the take.
REAPER Audio WorkflowsHow to record vocals in REAPER step by step (for beginners)
Just installed REAPER and cannot get your voice into it? Point REAPER at your audio interface, add a track, set it to your mono mic input, and record-arm it. Set the level on the interface so peaks reach about -12 to -6 dBFS, sort out monitoring, save the project, and press Record. Stock REAPER only, and a clean capture will not tune, de-room, or mix the take for you, but it makes every step after it far easier.
Home Vocal Recording and MixingVocal compression settings for beginners in REAPER (with stock ReaComp)
A vocal that jumps from too quiet to too loud needs compression, not more EQ. Add stock ReaComp, start around a 3:1 ratio, lower the threshold until the loudest phrases show 3 to 6 dB of gain reduction, use a medium attack and a quick or Auto release, then match the output level. Stock plugins only, and compression evens out level but will not fix tone, pitch, sibilance, or noise.
Podcast and Spoken-Word AudioHow to fix a remote guest's bad audio in REAPER (with stock plugins)
A remote guest often comes back thin, boxy, echoey, noisy, or quieter than you, and each of those is a separate fix. Get them on their own track, then work in order in REAPER: high-pass the rumble, reduce steady noise with ReaFIR Subtract, gate the room echo, restore presence with ReaEQ, and even the level with ReaComp. All stock, and it makes a rough call listenable but cannot undo internet codec damage, so capture a clean local recording next time.
Practical MixingVocal reverb settings that sound good in REAPER (with stock plugins)
Home vocals usually sound amateur because there is too much reverb, or it is dumped straight on the vocal track. Route ReaVerbate as a fully wet send, keep the pre-delay around 20 to 40 ms and the tail short, highpass the reverb near 200 to 300 Hz, and use less than you think. Stock plugins only, and reverb will not fix a muddy dry vocal or remove room echo.
REAPER Audio WorkflowsHow to comp vocal takes in REAPER (build one great take from many)
No single take is ever perfect, so comping builds one strong vocal from the best parts of several. In REAPER 7, set the track to Fixed Item Lanes, record a few passes, arm a comp lane, and swipe the best moments together. The classic show-all-takes-in-lanes method still works too. Stock features only, and comping picks between takes but will not fix pitch, tone, or a bad performance.
Home Vocal Recording and MixingHow to tune vocals in REAPER for free (with stock ReaTune)
REAPER ships with a free pitch corrector, ReaTune, so you do not need to buy anything to tune a vocal. Feed it a dry, mono lead, set the real key and scale so it stops snapping to wrong notes, and pick a vocal algorithm like elastique 2.1 SOLO. Use Automatic mode for quick cleanup or Manual to fix only the off notes, and know that tuning fixes pitch, not timing, tone, or a shaky take.
Audio Repair and RestorationHow to fix clipped and distorted audio in REAPER (and how much you can actually recover)
Clipping deletes the peaks of your waveform, so no plugin fully restores it. Check severity first: light, occasional clipping can be rebuilt with a declip tool like Audacity Clip Fix or iZotope RX De-clip, and REAPER's stock plugins can soften mild cases, but a brick-walled waveform is gone. Then prevent the next one with headroom.
Tools and RecommendationsThe best free vocal plugins in 2026 (and when REAPER's stock plugins already win)
REAPER's stock plugins already cover EQ, compression, de-essing, reverb, and pitch, so start there. Here is a short, honest list of genuinely free third-party plugins that add something real: TDR Nova, Klanghelm DC1A, Valhalla Supermassive, Auburn Sounds Graillon, and Voxengo SPAN, each verified free in 2026, plus what they will not fix.
Home Vocal Recording and MixingHow to fix thin vocals in REAPER (add body and air with stock plugins)
A thin vocal is missing low-mid body, or your own high-pass or compressor took it out. Bypass the chain to check the raw take, then add weight around 150 to 250 Hz and a touch of air with stock ReaEQ, and know what it will not fix. Stock plugins only.
REAPER Audio WorkflowsHow to set up a vocal template in REAPER (so every session starts ready)
Stop rebuilding your vocal setup every session. Save your track, FX chain, input, and routing as a REAPER template and load it in one step. Covers FX chains, track templates, and project templates, plus what a template will not do.
REAPER Audio WorkflowsHow to fix recording latency in REAPER (the delay in your headphones)
That echo or delay in your headphones while recording is monitoring latency. Lower it with a low-latency ASIO driver, a smaller block size, and hardware direct monitoring, and know what no buffer setting will fix.
Home Vocal Recording and MixingHow to de-ess vocals in REAPER (with stock plugins)
Harsh, whistling ess sounds are a level problem you can fix in REAPER for free. Find your sibilance frequency, then reduce it dynamically with ReaComp, ReaXcomp, or clip gain, and know what it will not fix. Stock plugins only.
Audio Repair and RestorationHow to reduce reverb (room echo) from a vocal in REAPER
Reverb is recorded into the same track as your voice, so stock REAPER can reduce it but not remove it. Use ReaGate as a gentle expander to tighten the pauses, trim room boom with ReaEQ, and know what it will not fix. Stock plugins only.
Podcast and Spoken-Word AudioHow to reduce background noise in a podcast in REAPER
Steady noise like a fan, AC, or hiss is reducible in REAPER for free. Capture a noise profile with ReaFIR, subtract it, add a gentle gate, and know what it will not fix. Stock plugins only.
Practical MixingVocal EQ cheat sheet for REAPER (frequency by frequency)
A frequency map for vocals: mud near 200 to 400 Hz, nasal near 1 to 2 kHz, presence and harshness in 1.5 to 5 kHz, sibilance in 5 to 9 kHz, air above 10 kHz. Starting points, not rules, with stock ReaEQ.
Audio Repair and RestorationHow to remove hum from a recording in REAPER
Mains hum sits at a fixed pitch, 60 Hz or 50 Hz depending on your region, plus harmonics. Here is how to notch it with ReaEQ or subtract it with ReaFIR using stock plugins.
Podcast and Spoken-Word AudioHow to get your podcast to -16 LUFS in REAPER (measure it, then hit it)
Most podcast apps expect about -16 LUFS integrated with true peak under -1 dBFS. Here is what LUFS means and how to measure and hit it in REAPER with stock tools.
REAPER Audio WorkflowsThe REAPER vocal chain with stock plugins (build order for beginners)
A complete vocal chain with REAPER's stock plugins: high-pass, corrective EQ, ReaComp, a de-esser, and reverb on a send. Here is the build order and why it matters.
Home Vocal Recording and MixingHow to fix harsh vocals in REAPER (with stock plugins)
A harsh, edgy vocal usually has too much upper-mid energy around 2 to 5 kHz. Here is how to find it with a sweep and tame it in REAPER with stock plugins.
Home Vocal Recording and MixingHow to fix nasal vocals in REAPER (with stock plugins)
A nasal, honky vocal is usually one loud resonance around 1 to 2 kHz. Here is how to find it with a sweep and reduce it in REAPER with stock plugins.
Home Vocal Recording and MixingWhy your home vocal sounds muddy in REAPER (and how to fix it)
Mud is a lack of clarity, and it is a tone problem you can fix without new gear. Here is how to find the low-mid buildup and clear it in REAPER with stock plugins.