Converting a Zoom recording to MP3

They have a meeting recording to send to a client, a transcriptionist or a colleague, and Zoom handed them a format the recipient does not want. Some of them only have the video file and do not realise the audio can be lifted out of it.

Zoom never writes an MP3. Its audio-only recording is always audio_only.m4a — AAC inside an MP4 container — and the video recording is zoom_0.mp4 with the same AAC track inside it. To get an MP3 you have to re-encode, because AAC is not legal inside an MP3 file and no amount of renaming makes it so. Find the file (local recordings live in Documents/Zoom, in a folder named after the meeting), drop it on the converter, choose MP3, and pick a bitrate no higher than the source had. For a meeting recording that usually means 64 or 96 kbps, not 320. The work happens in this browser tab; nothing is uploaded and no meeting audio leaves your machine.

Where Zoom puts the file

Local recordings — the only kind a free account can make — are written to the machine that hosted the meeting, after the meeting ends. The default folder is Documents\Zoom on Windows and ~/Documents/Zoom on a Mac, and Zoom makes one subfolder per meeting, named like 2026-08-24 10.00.00 Weekly sync 87654321098. Inside it you will typically find:

• zoom_0.mp4 — the video recording, H.264 video with an AAC audio track.
• audio_only.m4a — the audio-only recording, if that option was on.
• playback.m3u — a playlist pointing at the above, safe to ignore.
• chat.txt — the in-meeting chat.

If you cannot remember where the folder is, open the Zoom desktop client, go to Settings → Recording, and read the path next to Store my recordings at. The same panel has the checkbox that produces the audio-only file, and a second one, Record a separate audio file for each participant, which writes an Audio Record subfolder holding one .m4a per speaker. Those per-person files are gold for a transcript, because a transcription service that gets one speaker per file does not have to guess at diarisation.

Cloud recordings sit in the web portal instead, under Recordings → Cloud Recordings. Each meeting lists its parts separately, and Audio only is one of them. The download often arrives with a name like GMT20260824-100000_Recording.m4a, but it is the same AAC-in-M4A file. Cloud recording is a paid feature, so if you have never seen that tab, your recording is local.

Converting the M4A to MP3

Drop audio_only.m4a on the converter and choose MP3. The one decision that matters is the bitrate, and the honest answer is lower than most people reach for.

Zoom's audio-only file is speech, usually mono, and encoded at a modest bitrate to begin with. Asking for a 320 kbps MP3 from it produces a file five or six times the size with no extra detail in it, because the detail was never there. The converter reads the source bitrate off the file before you start and will tell you when the setting you picked is higher than the source, which is the moment to come back down.

Leave the channel setting alone unless the source is stereo and you do not need it. A mono source stays mono; forcing it to stereo doubles the size and writes the same signal twice.

This is a second lossy encode — AAC decoded, then compressed again as MP3 — and the converter says so on screen rather than pretending otherwise. For speech at these bitrates the second generation is close to inaudible, but it is not free, so do it once, from the original, rather than converting a file that somebody already converted.

If you only have the video recording

Plenty of hosts forget to tick the audio-only box and end up with nothing but zoom_0.mp4. The audio is still in there, and it can be lifted out without an encoder running at all: the track is already AAC, AAC is legal inside an .m4a, so the audio extractor copies the stream across and rebuilds the container around it. Every compressed byte survives, and because no sample is decoded the job takes about a second whatever the meeting's length.

If the recipient will take an .m4a, stop there — that is the best file you can hand them, and it is the one the transcription services generally prefer anyway. If they have insisted on MP3, extract to .m4a first and convert that, or convert straight from the MP4 in one pass. Either route re-encodes; the copy is only available when the target format can hold AAC.

One practical limit: an hour of screen-shared Zoom video is easily several hundred megabytes, and the converter runs inside a browser tab with roughly 2 GB of address space. It warns above 300 MB and refuses near the ceiling. If the MP4 is enormous, prefer the audio-only file, or ask the host to re-download just the audio part from the cloud portal.

Trimming the small talk

Meeting recordings start five minutes before the meeting does. Cutting that off before you send the file is worth doing for its own sake, and it makes every downstream limit easier to meet — a 62-minute recording that becomes a 48-minute one loses roughly a quarter of its bytes for free.

The audio cutter takes the start and end times and keeps what is between them. When the output format matches the input codec it copies rather than re-encodes, so trimming an .m4a to an .m4a leaves the audio exactly as Zoom wrote it. Trim first, convert second: cutting a finished MP3 to length is fine, but you will have burned the encode on material you then threw away.

Getting a long meeting under a limit

Two ceilings catch people with meeting recordings. Corporate mail systems commonly stop attachments somewhere around 20–25 MB, and the hosted transcription APIs enforce their own per-file size cap that a two-hour recording will sail past.

The fixes, in the order worth trying:

1. Trim. Free, and it removes material nobody wanted.
2. Go mono. A conference call is rarely genuinely stereo. Halving the channels roughly halves the bytes.
3. Drop the bitrate. 64 kbps mono is transparent enough for speech and for every speech-to-text model. Below about 32 kbps the artefacts start to cost you words in the transcript.
4. Split it. Two 45-minute halves clear a per-file cap that one 90-minute file cannot.

The arithmetic for the transcription case, including where the accuracy actually falls off, is worked through in getting an audio file under a transcription size limit. If you have not yet chosen a format for the recipient, which audio format transcription services want covers why M4A is usually fine and MP3 is only sometimes necessary.

Teams and Google Meet do the same job differently

Neither offers Zoom's audio-only file in the same place, so the video route is the normal one.

For the Teams and Meet files, download the MP4 and use the extractor. The AAC track copies out untouched exactly as it does from Zoom's MP4, which means the audio you send is the audio the platform recorded, not a second-hand version of it.

A note on where recordings end up

A meeting recording is usually the most sensitive audio file anyone converts. It has named people in it, sometimes a client, sometimes a salary. A cloud recording already sits with the meeting platform under the account administrator's control, and that is a decision your organisation has made. The conversion does not have to add a second copy somewhere else.

This converter is FFmpeg compiled to WebAssembly. It runs inside the tab, on your machine, and the file is read from disk by the page rather than sent anywhere. You can prove it: open your browser's developer tools, switch to the Network panel, and convert something. You will see the engine being fetched on the first visit and nothing outbound afterwards.

The other half of that is consent. Recording a meeting is a matter of law in some places and a matter of manners everywhere, and converting the file to a different format changes neither. If you are about to send a recording on to somebody who was not in the room, the format is the easy part of that decision.