Audio Repair and Restoration

Why your recording hisses: is it the mic or the interface?

By StudioRescue Editorial

Published September 30, 2026

Hiss is the most demoralising kind of noise problem, because it does not come and go like a passing truck. It is there in every take, it gets louder when you raise the track, and it tends to appear precisely when you try to make a quiet voice usable. The useful thing to know is that hiss is not one problem. It is three, they have different fixes, and you can find out which one you have in about five minutes without buying anything.

This article is about identifying the source of a broadband electronic noise floor. If your problem is a fan, traffic or a fridge, that is a different job and it is covered in how to reduce background noise in a podcast recording in REAPER.

The three places hiss comes from

1. The microphone’s own self-noise

A condenser microphone contains active electronics, and those electronics make a small amount of noise even in complete silence. Manufacturers publish this as equivalent noise level or self-noise. DPA notes that it is measured to the IEC 60268-4 microphone standard, and that two detector methods are in use: A-weighted RMS, and ITU-weighted quasi-peak. A-weighting reduces the low frequencies to better match how the ear responds at quiet levels.

That distinction matters the moment you compare two spec sheets, because DPA states the quasi-peak figure typically comes out 11–13 dB higher than the RMS figure for the same microphone. Most of the dB(A) numbers quoted in marketing are the A-weighted RMS ones, so check that you are comparing like with like before concluding one mic is noisier than another.

Neumann publishes a useful reading of what the numbers mean in practice: below 10 dB(A) is extremely low noise, 11–15 dB(A) is still very good, 16–19 dB(A) is good enough for most purposes, 20–23 dB(A) is a high figure where the noise becomes clearly audible, and 24 dB(A) and above is not really studio grade. If the spec sheet for your mic sits in that top band, its own noise may be your ceiling no matter what else you change.

Dynamic microphones are a different case. DPA points out that their self-noise is seldom specified at all, because the preamp usually determines the noise level. That is not because dynamics are quieter. It is because they are less sensitive, so they need more amplification, and the amplifier becomes the noisiest link.

2. The preamp working too hard

Preamps are rated by equivalent input noise, or EIN, which Sound Devices describes as the noise at the output minus the gain of the preamp. Removing gain from the number is what makes two preamps comparable. DPA gives a typical target range of about -120 dBu to -130 dBu, lower being better, and Sound Devices notes a hard theoretical floor of -133 dBV, which is simply the thermal noise of a 150 ohm source resistance.

Watch the units there, because dBV and dBu are not the same scale. A level in dBu sits about 2.2 dB above the same level written in dBV, so -133 dBV is roughly -131 dBu. In other words the best preamp physics allows sits only a decibel or so below the good end of DPA’s range, and two EIN figures are only really comparable when the measurement bandwidth, the weighting and the source impedance match. Spec sheets do not always say. Nothing, in any case, can be quieter than the noise the source impedance itself generates.

None of that matters while a loud source is hitting a sensitive microphone. It matters a great deal when a quiet source meets an insensitive microphone, because then you are asking the preamp for most of its available gain and its own noise arrives at a usable level alongside your voice.

3. The room and the machine

Air conditioning, a computer fan two feet from the mic, a hard drive, a nearby power supply. This is not electronic hiss, but it is broadband and steady, so it sounds like it. It also has the easiest fix of the three, which is why it is worth ruling out first.

The part almost everyone has backwards

The common belief is that turning the gain up adds hiss. That is not quite what happens, and the difference decides which fixes can possibly work.

Preamp gain raises the voice and everything already mixed into it by the same amount. Your microphone’s self-noise and your room’s noise both arrive at the input alongside the voice, so no gain setting improves the voice’s ratio against those. That part of the noise floor was settled at the capsule.

What adequate gain does do is lift the whole signal clear of the noise added after it, in the preamp’s output stage and the converter. That is the real reason the standard advice is to get your level from the preamp rather than from a fader afterwards. Sound Devices makes the related point that EIN is quoted with the gain subtracted out, so the noise you hear is independent of how much gain the preamp applies, a comparison that holds in the high-gain region the specification describes rather than at every setting on the knob.

Raising a fader or normalizing in REAPER after the fact is the case where nothing improves at all. The voice and every bit of noise already captured in the file go up together, by exactly the same amount.

Three recordings that tell you which one it is

Set your gain where you normally set it and leave it alone for all three clips. Record about ten seconds each.

Now compare. If clip one already hisses about as much as clip two, the interface is your limit and neither the mic nor the room is the problem. If clip two is clearly noisier than clip one, the noise is being added by the microphone, the room, or both, and you can separate those by switching off everything electrical in the room and recording clip two again. If clip two is genuinely quiet and the hiss only bothers you in clip three, the noise floor is fine and what you are hearing is a weak signal sitting too close to it.

Put a number on it

Ears get tired and monitoring levels drift, so it helps to read the three clips rather than only listen to them. REAPER can measure them for you. Cockos documents a loudness calculation performed by way of a dry run render, which reports statistics for the selected media, and REAPER’s normalization measurements include, among others, integrated LUFS, integrated RMS, peak and true peak. Run it on each of the three clips and write the numbers down.

Read the gap, not the absolute values. The distance between your voice clip and your silence clip is the signal-to-noise ratio you actually recorded, and it is the number that changes when you fix something. There is no published threshold that tells you when a home recording is clean enough, so make the final judgement by listening at the level your audience will use, not at the level you mix at. Hiss that vanishes on monitors at conversational volume can be obvious on earbuds.

The fixes, in the order that helps most

1. Get closer, or get louder. This is first because it is free and it is the only change that improves the ratio without touching the equipment. A louder source at the capsule means less gain is needed, so the voice rises while the noise floor stays where it is. Neumann makes the same point from the other direction: a low-noise microphone buys you freedom of placement, and a noisy one has to be worked close to get an acceptable signal-to-noise ratio. If you move in, watch for the low-end buildup and plosives that come with it.
2. Change the gain structure, not the gain amount. If you are near the top of the gain range just to reach a usable level, you are in the region where the preamp decides the result. An inline gain stage ahead of the interface, or a more sensitive microphone, moves that boundary. A different setting on the same knob does not. Sorting out where level is added is covered in how to gain stage vocals in REAPER.
3. Switch off what is making noise. Fans, heating, air conditioning, and where possible the computer. This costs nothing and is often the single largest number on the page.
4. Reduce what is left, last. ReaFIR ships with REAPER and is described by Cockos as an FFT-based dynamics processor that can, among other things, subtract noise from a signal, and ReaGate can hold the gaps down between phrases. Both are remediation rather than repair. The method is in the background noise article, and if you want to compare dedicated tools first, see the best free noise reduction plugins for voice.
5. Do not stack it. Every layered take carries its own copy of the noise floor, so doubles and harmony stacks raise it even when each take sounds acceptable alone. Because those copies are unrelated to one another they add up by roughly 3 dB each time you double the number of tracks, rather than piling on one for one. That is gentler than it sounds, but it still accumulates across a large stack. Lewitt makes the same point and adds that compressing afterwards is what makes it obvious. Fix the noise floor before you build layers on it.

What noise reduction will not fix

• It cannot give back what the noise was covering. Breaths, room tail and quiet consonant detail sit in the same range as the hiss, so aggressive reduction takes them with it and the voice starts to sound processed and underwater.
• It will not rescue a take that was recorded far too quietly and then raised, because raising the take raised the noise floor with it. That case is covered in how to fix a vocal recorded too quietly in REAPER.
• It does nothing about the next take. A mic with a high self-noise figure will produce exactly the same floor tomorrow.
• It cannot reliably separate hiss from hissy sounds in the voice itself. Sibilance lives in the same territory, and a noise reducer aimed at one will usually dull the other.

If it still hisses

1. Check the boring hardware first. A different cable, a different input on the same interface, and phantom power confirmed on for a condenser. A failing cable or an input set to the wrong type can raise a floor all by itself.
2. Make sure you are judging the file, not the monitoring. Headphone amps and monitor controllers have their own noise floors. Look at the recorded clip rather than what you hear while the session is idle.
3. Check your monitoring level. If you mix loud, you will hear a floor your listeners never notice. If you mix quiet, the opposite. Decide at the listener’s level.
4. Accept the diagnosis if clip one is the noisiest thing in the chain. This is the one case where hardware genuinely is the answer, and it is the interface rather than the mic. Before spending anything on a microphone, read will a better microphone fix my vocals, because most tone complaints are not solved by buying, and hiss is one of the few that sometimes is.