Can an audio enhancer fix a bad microphone in real time?

The call ends, you download the recording, drop it into Adobe Podcast's Enhance Speech, and a few minutes later you have a cleaned-up file. Then the uncomfortable thought arrives: the eight people on the call never heard that version. They heard the raw MacBook mic, the fan, the kitchen echo. The enhancer fixed the copy nobody will play again.

That gap is the whole question. "Audio enhancer" covers two different kinds of software, and most of the famous ones only work on files. A smaller group works live, between your mic and Zoom, and it plays by stricter rules. This article separates the two, says what each can and cannot repair, and ends with a setup you can try today.

The short answer

Yes, partly. A real-time audio enhancer can remove steady background noise, cut some room echo, even out your level and reshape the tone of a thin or boxy mic, all while you talk. It cannot invent detail the microphone never captured, and it cannot use the trick that makes file-based tools so strong: hearing the whole recording before deciding what to do with any part of it.

Think of it this way. A file enhancer is a restorer working on a finished painting with unlimited time. A live enhancer is a sound engineer at a concert desk: every decision has to be made now, with no way to undo it, and the audience hears the mistakes.

If your only problem is a dog, a fan or a keyboard, you may not need an "enhancer" at all. That is a noise suppression job, and the comparison of noise cancelling software that works on calls is the better page for you. Stay here if your voice itself sounds wrong: thin, distant, muffled, too quiet, harsh.

Two jobs hiding behind one search term

People typing "audio enhancer" or "audio enhancement software" are usually looking for one of these:

The word "enhancer" makes them sound interchangeable. They are not. You cannot upload a Zoom call to Adobe while it is happening, and a live filter tuned not to glitch will never be as aggressive as a tool that gets to think about a file for two minutes.

Why file enhancers can do more

Adobe describes Enhance Speech as a tool that "fixes audio problems after the fact." The phrase "after the fact" carries most of the weight.

When a tool has the whole file, it can analyze the room tone across all thirty minutes, find the quietest gaps to learn what the noise is, compare the start of a word with its end, and run heavy models with no deadline. If one pass makes a mistake, the user re-runs it. Auphonic's leveler can look at the whole episode before deciding how loud each sentence should be. Audacity's Noise Reduction asks you to select a stretch of pure noise first ("Get Noise Profile") and only then apply the reduction to the rest.

A live processor has none of that. The sound of the word you are saying right now has to be cleaned, shaped and handed to Zoom before you finish the sentence. It has seen the last few seconds at most. If it buffers too much, you start talking over people because your voice arrives late. If it guesses wrong about what is noise, the guess goes straight to the call.

This is why live tools are built to be conservative, and why the ones that try to be clever tend to be expensive in CPU. Zoom's own help page warns that the High level of background noise suppression "may increase CPU usage." Discord says its Krisp filter turns itself off when CPU usage gets very high. Yandex Telemost, the meeting app many Russian teams use, documents the same behavior. A file enhancer that runs slowly is an annoyance. A live enhancer that runs slowly breaks the call.

The best free audio enhancer for a finished recording

If the recording is already made, use a file tool. It will do more than any live tool can. As of October 2026, the free options worth knowing:

Which one wins depends on the job. For a single painful file, such as an interview recorded on a laptop in a café, Adobe's free Enhance Speech is the obvious first try, because rebuilding speech is exactly what it was made for. For a weekly podcast, Auphonic is the better habit: two free hours a month covers a short show, and loudness targets matter more over a season than one dramatic rescue. For a hum that never changes, Audacity costs nothing and gives you full control.

None of these help with tomorrow's meeting.

Real-time options on a Mac, free ones first

Search for a "mic enhancer", a "microphone enhancer" or a "voice enhancer app" for Mac and most of what comes back is noise filtering under a new name. Before paying for any audio enhancer app, look at what is already on your Mac and inside your apps.

macOS Mic Modes. On macOS Sonoma 14 or later, during a call in an app that supports it (FaceTime is Apple's example), click the Video or Audio item in the menu bar and choose Mic Mode. Voice Isolation, in Apple's words, "minimizes the sounds of traffic, dogs barking, and other background noise." Wide Spectrum keeps the room in. Standard turns the processing off. Apple notes the options vary by app, so check whether your call app shows the menu at all.

Zoom. Settings > Audio offers Microphone modes: Noise removal (the default), Personalized audio isolation, and Original sound for musicians. Under Noise removal, Background noise suppression runs at Auto by default, with Low, Medium and High. High removes background speech too, and Zoom warns it may raise CPU usage. Original sound for musicians switches the post-processing off, which is what you want when something earlier in the chain already did the cleaning.

Discord. User Settings > Voice & Video > Noise Suppression gives you Krisp for free inside Discord. On a Mac it needs Apple silicon and macOS 12 or later. It only filters your side. Discord itself suggests leaving it off in a quiet room, which is a useful hint about what noise suppression costs: when there is nothing to remove, the filter can only take away from your voice.

OBS. For streams and recordings, the Noise Suppression filter offers RNNoise (the default, better quality, more CPU) and Speex (adjustable, default -30 dB). OBS's own documentation says the filter is not effective against large amounts of noise in a loud room.

Krisp. A standalone app that adds a "Krisp Microphone" you select in any call app. As of October 2026 there is a 7-day free trial, then Core at $16 per user per month, or $8 billed annually, and noise cancellation comes inside every plan, bundled with other meeting features such as accent conversion. The Krisp review for Mac users goes through whether that bundle makes sense.

NVIDIA Broadcast. Free, with Noise Removal, Room Echo Removal and Studio Voice, but it needs Windows and a GeForce RTX 2060 class card or better. There is no Mac version; the NVIDIA Broadcast for Mac page covers what people use instead.

Almost every tool in this list is, at its core, a subtractor. It decides which part of the signal is not your voice and turns it down. That is valuable, and for a noisy room it may be all you need. But it explains why a bad microphone often still sounds bad after you switch on Voice Isolation.

What a "bad mic" usually means, and what fixes each part

When someone says their mic sounds bad, noise is only one of the seven complaints in the table below. Most of the others are not about extra sound. They are about the voice itself being shaped wrong.

Look at the third column. Four of the seven fixes are tone shaping or level control, which noise suppressors do not attempt. Read the audio settings of Zoom, Discord or Telemost: the processing options there remove sound (noise, echo, background voices) or at most ride the level, and none of them offers a control for the tone of your voice. A setting that has to suit every voice on every mic cannot shape tone for yours.

Two of the rows are not software problems. Distance comes first: sound level drops by about 6 dB every time the distance doubles, so halving the gap between mouth and mic lifts your voice against the room before any software runs. And Bluetooth headsets in call mode usually carry the mic over the Hands-Free Profile, which samples at 8 or 16 kHz, so much of the detail is gone before any enhancer sees it. Before buying any software, work through the free fixes in how to make your mic sound better without buying a new one. Some readers will stop there.

Where TunedMic fits

TunedMic is a paid Mac app built for the rows of that table a live tool can handle, and specifically for the tone and level rows that noise filters leave alone.

It adds a virtual microphone called TunedMic to your Mac. Your real mic stays where it is; the app takes its sound, processes it on the Mac in real time, and any app where you can choose a microphone can pick it: Zoom, Google Meet, Microsoft Teams, Slack, Discord, OBS, FaceTime, browsers. If the idea of a second, software microphone is new to you, what a virtual microphone is and why a Mac would need one explains the mechanism.

The difference from a generic enhancer is the setup. Instead of applying a preset made for an average voice, the app runs a Sound Check: you read a short passage aloud for about a minute, in English, Spanish, German, French, Chinese or Russian, and it takes more than 30 measurements of your voice, your mic and your room. The nine profiles (Natural, Warm and Focused for calls; Balanced, Broadcast and Crisp for creators; Gentle, Open and Forward for singing) are built from that measurement rather than from a stock curve. Manual controls for noise, room echo, clarity, loudness and a 12-band equalizer exist if you want them, and you never have to touch them.

The real-time constraints from earlier apply here too, and the app states its numbers. It adds less than a tenth of a second of delay, which on a call disappears into the connection's own delay. Processing runs locally, and the voice is never uploaded. The limits are just as specific:

• It needs a Mac with Apple silicon (M1 or newer) and macOS 14 or later. No Intel, no Windows, no iPhone.
• There is no free tier. You get 7 days free with the full app, then it is $19 once, with a year of updates and a license for two Macs.
• It keeps you sounding like you. It is not a voice changer.
• It cannot rebuild what the mic never captured. A Bluetooth headset gives it less to work with than a wired or USB mic.

That last point is why your own mic is the real test, and the 7 free days are there for it. Before you start them, the 38-second video on the home page shows what setup asks of you, one minute of reading aloud, and the Speech Report that arrives after a call. Judge the result on your own hardware, rather than on a description.

Picking the right tool for your situation

Do not stack two noise suppressors

One mistake undoes most of the gains: running two noise filters in a row. Each one guesses what is noise. The second guesses based on what the first already changed, and it tends to bite into the voice, because the remaining "noise" is mostly the edges of your words.

If you use a processor before the call app, whether it is Krisp, Voice Isolation or TunedMic, set the call app's own filter low or off. A chain that keeps each job in one place looks like this:

Real mic (hand's width from mouth, input level peaking well below the red)
  -> one live processor (noise, echo, tone, level)
  -> call app
       Zoom:    Background noise suppression = Low
                (or Original sound for musicians if the processor already cleans noise)
       Discord: Noise Suppression off
       OBS:     no second noise suppression filter on the same source

The same logic applies to recordings. If you process live and then run the file through Adobe Enhance Speech as well, the second pass will treat the first pass's work as raw material. Pick one layer for each job.

Try this before your next call

Open QuickTime Player, choose File > New Audio Recording, and record thirty seconds of yourself talking at your normal call distance. Play it back on headphones and name the problem in one word: noise, echo, thin, muffled, quiet or harsh.

If the word is "noise", switch on your call app's suppression and you are done. If it is anything else, move the mic closer and fix the input level, then record again. If the voice still sounds wrong and you are on a Mac with Apple silicon, try TunedMic on your own mic during the free 7-day trial, and if it fixes the problem you named, the one-time price is on the pricing page.