What is Audio MIDI Setup on a Mac, and which mic settings in it matter?

A forum reply says the fix for your thin Zoom voice is simple: open Audio MIDI Setup, set the microphone to 96 kHz and 24-bit, done. You find the app, select your USB mic, open the Format menu and pick the biggest numbers on the list. The next call sounds exactly like the last one. Twenty minutes later the recording app starts to crackle: the mic now runs at 96 kHz and the headphones are still at 48.

That is the usual first meeting with Audio MIDI Setup on a Mac. It looks like a mixing desk. It works more like the fuse box in a hallway: it shows how every audio device is wired and lets you rewire a few things, but none of its switches makes a voice clearer. A few of its fields do matter for a microphone, mostly because a wrong value there breaks things without any warning. The rest are best left exactly as macOS set them.

If you only want to choose which mic the Mac uses, raise its volume or give Zoom permission to hear it, you do not need this app at all. How to change microphone settings on a Mac covers those everyday controls in System Settings, and it is the better page for that reader. This one is for the person staring at the Format menu, the Clock Source pop-up or the plus button and wondering what each does.

What Audio MIDI Setup on a Mac actually is

Audio MIDI Setup is a utility Apple ships with every Mac. It has two halves, and the name advertises both.

• The Audio Devices window lists every audio input and output macOS can see: the built-in microphone and speakers, a USB mic, an audio interface, AirPods, a display with speakers, and virtual devices installed by software such as BlackHole or Loopback. For each one it shows the format the device is running at right now and lets you change it.
• MIDI Studio, opened from Window > Show MIDI Studio, is for keyboards, controllers and other MIDI gear. MIDI carries notes and knob movements, not sound. For a microphone you can ignore this half entirely.

What the app does not have matters as much. There is no equalizer, no noise reduction, no compressor, no "voice enhance" checkbox. Audio MIDI Setup is plumbing. It decides which pipe connects to which and how wide the pipe is. What flows through it is set somewhere else.

How to open Audio MIDI Setup

Three routes, all to the same window:

1. Finder. Choose Go > Utilities, then open Audio MIDI Setup. Apple documents this path. On disk it lives in Applications > Utilities.
2. Spotlight. Press Command and Space, type "audio midi", press Return.
3. Terminal. open -a "Audio MIDI Setup" launches it directly.

If the window that appears shows MIDI icons instead of a device list, choose Window > Show Audio Devices, Apple's documented path back.

There is also a read-only shortcut that never risks changing anything. In Terminal, system_profiler SPAudioDataType prints every device with its channel count and current sample rate. Here is a trimmed example from an Apple silicon MacBook Pro running macOS 26 with a USB wireless receiver plugged in:

MacBook Pro Microphone:
  Input Channels: 1
  Current SampleRate: 48000
  Transport: Built-in

Wireless microphone:
  Input Channels: 2
  Current SampleRate: 48000
  Transport: USB

MacBook Pro Speakers:
  Output Channels: 2
  Current SampleRate: 48000

Two facts jump out before you touch a single setting. The built-in mic is a one-channel device. The wireless receiver reports two channels, which becomes important a few sections down.

Which Audio MIDI Setup fields matter for a microphone

Select your mic in the sidebar on the left. The right side fills with controls. Here is the full set, with a verdict for each one if your goal is calls, streams or voice recording.

So the honest answer to "which mic settings in Audio MIDI Setup matter" is two: Format and the volume sliders. Everything else is either a duplicate of System Settings, a control the Mac already sets correctly, or something for speakers.

Format: sample rate and bit depth

The Format pop-up sets sample rate and bit depth in one choice. Apple's own instruction is short: choose the settings appropriate for the device. In practice that means one rule and one number.

The rule: every device in the chain should agree. If the mic runs at 48 kHz and your recording app or headphones run at 44.1 kHz, something has to convert between them, and a conversion step inside an app is a common source of crackles and drifting timing. Audio-Technica's manual for the AT2020USB-X spells it out for its own mic: match the sample rate to the value set in the recording software, because a mismatch may produce noise. The same mic even signals the problem physically, with its mute lamp blinking red when the rates differ.

The number: 48 kHz. Audio-Technica recommends it for stable recording with less load on the computer, even though that mic offers up to 96 kHz. In the example above, the built-in mic, the wireless receiver and the speakers already run at 48 kHz, so choosing it for an external mic usually means nothing needs converting.

Why not the biggest number? A sample rate captures frequencies up to half its value. At 44.1 kHz that ceiling is 22.05 kHz, at 48 kHz it is 24 kHz, both above the range of human hearing and far above where a speaking voice carries its energy. 96 kHz doubles file size and processing for frequencies no listener on a call will ever receive, since the call app encodes your voice for the network anyway. The full reasoning, including when 24-bit beats 16-bit, is in which sample rate and bit depth to use for voice.

Bit depth is simpler. If the menu offers 24-bit, take it: you can set the level a little lower, further from clipping, without raising the noise floor. It will not make a voice sound different on Zoom.

One quirk catches people every time. Some USB mics with a headphone jack appear twice in the sidebar, once as an input and once as an output. Audio-Technica tells AT2020USB-X owners to pick the entry that shows "1 in / 0 outs" when setting the mic's format. If you change the format and nothing seems to happen, check that you edited the input entry.

Input volume sliders

Under the Input tab each channel gets a volume slider. Apple's guide notes that the sliders are blue only when the device supports software volume control. Blue means you can set the gain here, per channel. Grey means the device keeps its gain in hardware, usually on a knob on the mic body, and nothing on the Mac will change it. The same limit is the likely reason the Input volume slider in System Settings does nothing for some USB mics.

When the sliders are blue, the job is the same as with any gain control: speak at your normal call volume and keep the loudest syllables clear of the top. A signal that hits the ceiling clips, and no app can remove clipping afterwards. What gain staging is and how to stop your mic from clipping walks through setting it in the right order, mic knob first, Mac second, app last.

Clock Source and Hardware Rate Converter

Apple's own wording for Clock Source is that it is generally set to the default and usually needs no adjustment. It matters in studios where several digital devices share one external clock. On a single USB mic there is nothing to synchronise, so leave it.

Hardware Rate Converter appears in the Input pane only on compatible Macs, with Automatic, Enable and Disable. Automatic is the right answer unless a device manufacturer tells you otherwise.

Two channels, one voice: the left-ear problem

Back to that wireless receiver reporting two input channels. Many USB devices, audio interfaces especially, present themselves as stereo even when a single mic is plugged in. The mic lands on channel 1, channel 2 carries silence or a second transmitter, and what happens next depends on the app, not on Audio MIDI Setup.

Focusrite documents two cases on macOS. Keynote and QuickTime accept only stereo input and assign Input 1 to the left channel, so a mic on Input 1 is recorded and played back on the left side only. Discord, on a Mac, receives only Input 1 of an interface, so a mic plugged into Input 2 may not be heard at all.

The fixes live outside Audio MIDI Setup:

• Record in an app that lets you choose a mono input, such as GarageBand, instead of QuickTime.
• On an interface, plug the voice mic into Input 1.
• In OBS, Focusrite's instructions for a Mac are to untick Enable Downmixing and set Channel 1 and Channel 2 to the same mic input, since every OBS source is stereo.

Focusrite also mentions System Settings > Accessibility > Audio > Play stereo audio as mono. That changes only what you hear on your own Mac. The recording stays one-sided, so treat it as a way to listen, not a repair.

AirPods and Bluetooth: the number you cannot change

Select AirPods in Audio MIDI Setup while a call is running and the Format may show a rate far below 48 kHz. Jeff Geerling documented AirPods stuck at 16 kHz in voice mode and returning to 48 kHz after the mic was released. A December 2024 report on Apple's forums describes AirPods 4 on macOS Sequoia dropping to 24 kHz whenever the mic is in use, with no way to switch back in Audio MIDI Setup.

That is not a setting to fix. Bluetooth narrows the link as soon as the headset's mic is active, and the Format menu simply reports the result. Changing it does nothing useful. If the AirPods mic sounds poor on a Mac, the practical move is a different input: keep the AirPods for listening and pick the built-in mic or a USB mic in Input. Why your AirPods mic sounds bad on a Mac covers the rest.

Aggregate devices: when one mic is not enough

The plus button at the bottom of the sidebar offers Create Aggregate Device. An aggregate glues several devices into one, so an app that accepts a single input device can see the channels of two or three. Apple's own example is a podcast in GarageBand with two USB mics, recorded to separate tracks.

The setup, following Apple's steps:

1. Click the plus button and choose Create Aggregate Device. Double-click the new entry to give it a name you will recognize later.
2. Tick Use for each mic you want inside it. The order you tick them in sets the order of the channels.
3. Set every included device to the same sample rate first. Apple says so directly: the aggregate works correctly only when the rates agree.
4. Choose the Clock Source: Apple's advice is the device with the most reliable clock.
5. Turn on Drift Correction for every device that is not the clock source. Two USB mics each keep their own time, and without correction their tracks slowly slide apart over a long recording.
6. In GarageBand, go to Settings > Audio/MIDI, pick the aggregate as the input device, then give each track its own input channel. Dragging devices left or right in Audio MIDI Setup changes which channel numbers they get.

The BlackHole documentation adds one practical warning: an AirPods mic runs at a lower sample rate and should not be the clock device in an aggregate.

What an aggregate will not do is mix. On a Zoom or Telegram call, you still pick one input, and a call app expects one voice. If two people need to share one call input, or you want a mic and a music track blended before they reach the app, that is a routing job for tools like Loopback. Do you need Audio Hijack or Loopback explains where each one fits.

Multi-output devices: the half people confuse with mic settings

The same plus button offers Create Multi-Output Device. This one is about sound going out, not coming in: audio sent to it plays through every device you tick, all at once. Its most common job is recording the Mac's own audio with BlackHole while still hearing it in your headphones.

Two traps, both from the BlackHole documentation: macOS cannot change the volume of a multi-output device, so the usual volume control does nothing while it is selected, and you set each device's level in Audio MIDI Setup instead. And drift correction belongs on every device except the clock source, or the audio glitches after a while. Whether BlackHole alone covers your recording or you need Loopback is answered in Is BlackHole enough to record your Mac's internal audio.

What Audio MIDI Setup cannot change about your voice

Here is the part the 96 kHz advice skips. Once the format matches and the gain is clean, Audio MIDI Setup has nothing left to offer your voice. Thin, boomy, echoey, too quiet next to the other people on the call, buried under a fan: none of these is a format problem.

What changes the sound, in order of cost:

• Placement. Halving the distance to the mic raises your voice against the room. Free, and stronger than any slider.
• Mic Mode. On macOS Sonoma and later, during a video call, the Video or Audio menu in the menu bar offers Mic Mode. Voice Isolation gives priority to your voice and pushes down traffic, barking and other background noise. It applies only in apps that support it and only while a call is running.
• The call app's own processing. Zoom, Discord and Telegram Desktop each keep their own mic selection, and Zoom and Discord add noise suppression of their own. None of them share settings with each other. How good Discord noise suppression is shows where that built-in layer runs out.
• Processing that sits between the mic and every app. On a Mac, that is the slot TunedMic is built for. It adds one more microphone, called TunedMic, that any app can pick, from Zoom and Google Meet to Telegram, Discord and OBS. Your real mic stays as it is. A one-minute Sound Check, read aloud in one of six languages, takes more than 30 measurements of your voice, your mic and your room, and nine profiles are built from that. Noise, room echo, clarity, loudness and a 12-band equalizer are there if you want to adjust by hand. You never have to touch them.

The limits are real. TunedMic needs a Mac with Apple silicon and macOS 14 or later; Intel Macs and Windows are not supported. With AirPods or any Bluetooth headset it works, but Bluetooth limits how much detail the mic sends, so a wired or USB mic gives it more to work with. And it does not replace the format fix above: a mic running at a mismatched rate should be fixed in Audio MIDI Setup first.

A settings sheet for one mic on calls and streams

If you want one screen to copy, this is it. Values are for a single USB mic or interface used for calls, streams and voice recording.

Audio MIDI Setup > select your mic > Input tab
  Format ................ 48,000 Hz, 24-bit if offered
  Volume sliders ........ blue: set so loud syllables stay clear of the top
                          grey: set gain on the mic itself
  Clock Source .......... Default
  Hardware Rate Conv. ... Automatic

Audio MIDI Setup > select your headphones or speakers > Output tab
  Format ................ 48,000 Hz (same as the mic)

Recording app or OBS
  Sample rate ........... 48 kHz (same again)

Leave alone unless you need them
  Aggregate Device ...... two or more mics into one recording app
  Multi-Output Device ... hearing audio while BlackHole records it

When Audio MIDI Setup itself looks wrong

The mic is missing from the list. Apple's checklist for a device that is not working: it is connected properly, it is switched on, and any software the manufacturer requires is installed. Many USB mics need no driver at all; whether your USB mic needs its own software sorts out which ones do. If the mic is listed but no app hears it, the cause is usually permissions or app settings, covered in why your Mac microphone is not working.

Audio apps stopped working after you installed a driver. Apple's advice is to get the latest driver from the manufacturer, and if no compatible one exists, return to the standard macOS drivers: select the Mac's own speakers and choose Use This Device for Sound Output from the Configure Selected Device menu.

The format keeps changing back. Recording apps can set the device rate themselves. Apple's Logic Pro documentation says an interface's rate should adjust automatically to match the project, which you check in File > Project Settings > Audio. If the format jumps to 44.1 kHz every time you open a project, change the project rate rather than fighting the menu.

The next step

Open Audio MIDI Setup now, select your mic and look at two things: the Format, and whether the volume sliders are blue. Set the mic and your headphones to the same rate, 48 kHz unless a recording app needs otherwise, set the gain if the sliders let you, then close the app and leave the rest alone.

Then make a short test call or recording. If the voice is still thin, echoey or noisy, the remaining problem is the sound itself, not the plumbing. The 38-second video on the TunedMic home page shows what the fix involves: Sound Check, a minute of reading aloud, after which the mic is tuned to your voice for Zoom, Google Meet and Discord. If tuning is the change you need, the free 7-day trial starts at first launch and includes every feature with no card or account. Payments are final and there are no refunds, so use those seven days on your own calls before you look at pricing for the one-time license.