Practical Mixing
How loud should vocals be in the mix? (and how to check it in REAPER)
Published September 19, 2026
There is no universal decibel number for vocal level, and any specific figure you see quoted online is somebody’s personal taste presented as a rule. What decides it is three checks: compare your balance against a reference track in the same style, listen quietly, and listen in mono. If the vocal survives all three, the level is right. Then use REAPER’s stock loudness metering to write down the difference between your vocal and your instrumental so you can repeat it on the next song instead of starting from zero every time.
This is one of the most searched questions in home recording, and most answers either give you a made-up number or tell you to use your ears and leave it there. Neither helps. The useful version sits in between: you do decide by ear, but you can measure the decision afterwards so it stops being a coin flip.
One thing to be clear about first. This article is about mix balance, which is how loud the vocal sits against everything else. It is not about setting your recording level in REAPER, which happens before you ever open the mixer, and it is not about vocal tone. Those are separate jobs with separate fixes.
Why no single number exists
Three reasons, and all three are worth understanding because they tell you what to do instead.
- The platform sets the final loudness, not you. Streaming services normalize playback. Spotify, for example, adjusts tracks to
-14 dB LUFSusing the ITU 1770 measurement standard, and its listeners can pick quieter or louder playback settings on top of that. So the absolute loudness of your export gets moved anyway. What survives normalization is the relationship between your vocal and your instrumental, which is the only thing you actually control. - Your ears change the balance as you change the volume. Human hearing is not equally sensitive at all frequencies, and that sensitivity shifts with level. This is what the equal-loudness-level contours in ISO 226 document. In practice, a mix that sounds balanced loud can sound vocal-heavy quiet, or the reverse. A single number cannot survive that.
- Style conventions differ, a lot. In most modern pop and rap productions the lead vocal is conventionally the most prominent single element in the mix. In a dense rock or metal mix it usually sits further back and more embedded. Neither is more correct. A number that fits one would be wrong for the other.
You will find articles confidently stating that vocals should sit “3 to 6 dB above the instrumental” or similar. No standards body and no manufacturer publishes a figure like that. Treat those numbers as one person’s preference, not as a spec.
The three checks that actually decide it
Set a level that sounds right to you first, then run these. They are quick, and they catch almost every balance mistake.
The reference check
Import a released track in the same style into your REAPER project on its own track. Drop its fader until it is roughly as loud as your mix, then switch back and forth. Do not listen for tone. Listen only for how far forward the vocal sits. Home mixes usually turn out to have the vocal too low on this test, because you have heard your own words a thousand times and your brain fills them in even when they are buried.
The quiet check
Turn your monitoring down until the mix is barely comfortable, about the level of a background conversation. At low volume the loudest element is the one you still hear clearly. If your vocal vanishes into the beat, it is too low. If everything else vanishes and the vocal is left floating alone, it is too high.
The mono check
Collapse the mix to mono and listen again. Wide stereo instruments lose level in mono while a centred vocal does not, so a balance that only works in stereo will show up here. Phones, laptop speakers and many smart speakers play something close to mono, so this is not a theoretical test.
Turn your choice into a number you can repeat
This is the part almost nobody does, and it is the reason the question comes back on every song. Once your ears have picked a balance, measure it. Then next time you have a starting point instead of a blank fader.
1. Get your mix onto two buses
Route all the vocal tracks to one bus and everything else to another. If you have not set that up before, how to use a vocal bus in REAPER walks through the routing. Two buses is the minimum you need to compare anything.
2. Measure each bus
REAPER gives you two stock ways to read loudness, and you do not need to install anything.
- The JSFX loudness meter. REAPER ships with a JSFX loudness meter that reads LUFS. Put one on the vocal bus and one on the instrumental bus, play a busy section of the song, and note the short-term readings each one settles around.
- A dry run render. REAPER’s render dialog can perform a fast dry run that produces a loudness and clipping report without writing any media to disk, and it works on stem renders. Run it once per bus and read the figures off the report.
3. Write down the difference, not the values
The individual numbers will change the moment you touch the master fader. The gap between them will not. Note something like “vocal bus reads about four LUFS higher than the instrumental bus on the chorus” in your project notes or your REAPER vocal template. That figure is yours, derived from a balance you approved by ear, and it is a far better starting point on the next song than any number from an article.
You are not outsourcing the decision to a meter. You are recording a decision you already made so that you do not have to make it from scratch again. The ear picks the target, the meter remembers it.
If you keep reaching for the fader, the problem is not the level
A very common pattern: the vocal sounds right in the verse, buried in the chorus, and too loud on one line. People respond by hunting for a better fader position. There is no better fader position, because the problem is consistency, not balance.
- The performance moves more than the mix can absorb. Even out the dynamics first with vocal compression settings for beginners, then set one balance for the whole song.
- A few specific words are the problem, not the section. That is a job for volume envelopes. See how to automate vocal volume in REAPER.
- The vocal is being masked, not buried. If the vocal disappears only when a thick synth or guitar enters, the two are fighting for the same frequency range. Raising the vocal makes it louder without making it clearer. Clearing space in the other part usually works better, and why your home vocal sounds muddy in REAPER covers the low-mid range where this happens most.
- It is speech over music. For a podcast or voiceover, a static balance is the wrong tool entirely. Use how to duck music under voice in REAPER instead.
| What you hear | Likely cause | Where to fix it |
|---|---|---|
| Vocal disappears at low listening volume | Balance genuinely too low | Raise the vocal bus, then re-run the reference check |
| Vocal sounds detached, sitting on top of the track | Balance too high, or no reverb tying it in | Lower it a little and judge against a reference |
| Right in verse, buried in chorus | Performance dynamics | Compression, then one static balance |
| Only certain words vanish | Word-level level swings | Volume envelopes |
| Fine in stereo, buried in mono | Stereo width hiding the issue | Set the balance with the mono check applied |
| Loud enough but still unclear | Frequency masking | Make room in the competing part with EQ |
What setting the right level will not do
Balance is one variable. It is worth being blunt about what moving a fader cannot reach.
- It will not fix tone. A muddy, harsh or nasal vocal is still muddy, harsh or nasal at any level. Those are EQ problems.
- It will not fix a noisy or roomy take. Turning a noisy vocal up turns the room and the hiss up with it, in exactly the same proportion.
- It will not repair a take that clipped on the way in. That damage is in the file, not in the mix.
- It will not make an uneven performance sit still. No single fader position covers a take that swings widely in level.
- It will not make your track competitively loud. Overall loudness is a mastering question and it is separate from where the vocal sits inside the mix.
Getting the balance right is the difference between a mix that sounds amateur and one that sounds finished, and it costs nothing. It is also the single change most likely to make a listener say your mix improved, which is why it is worth measuring rather than guessing.
Stop looking for a dB number, because there is not one. Set the balance by ear, then confirm it against a reference track, at low volume, and in mono. Once it passes all three, measure the gap between your vocal bus and your instrumental bus with REAPER’s stock loudness metering and write that gap down. That is your number, and it is the only one worth trusting.
Want a consistent vocal before you set the balance?
A vocal that swings in level makes balance impossible, because no single fader position fits it. The free StudioRescue vocal chain measures your take and sets REAPER’s stock plugins to even it out first. It does not set the balance for you, and it is not meant to. It just gives you something steady to balance.
Get the free vocal chainSources
- Cockos, REAPER (version 7.80, released September 13, 2026) - per-track LUFS loudness metering, the stock JSFX LUFS loudness meter, and render statistics including fast dry run renders that produce a loudness report without writing media to disk. reaper.fm
- Cockos, REAPER User Guide (updated for version 7.80). reaper.fm/userguide
- Spotify for Artists, “Loudness normalization on Spotify” - tracks adjusted to -14 dB LUFS according to the ITU 1770 standard, with listener-selectable Loud, Normal and Quiet playback levels and a recommended true peak ceiling. support.spotify.com
- ITU-R Recommendation BS.1770 - algorithms to measure audio programme loudness and true-peak audio level, the measurement standard behind LUFS. itu.int
- ISO 226, Acoustics: normal equal-loudness-level contours - the standard documenting how perceived loudness varies with frequency and level. iso.org