Lossless, lossy, and the one-way door
WAV stores the sound as plain numbers, one per sample, with nothing thrown away. Open it, edit it, save it, repeat a hundred times, and the hundredth copy is identical to the first. The price is size: around 10 MB per minute of stereo at CD quality.
MP3 and M4A are lossy. They discard the parts of the sound a model of human hearing says you will not miss, and get the file down to a tenth of the size or less. For a podcast or a voice memo the loss is inaudible. For music at a low bitrate it is not.
The rule that follows from this is the only one worth memorising. Lossless to lossy is fine. Lossy to lossy is a loss on top of a loss, and lossy to lossless gives you a big file containing a small file's quality, which is the worst of both. Keep the lossless original, make lossy copies from it, and never make a lossy copy from a lossy copy if you can avoid it.
M4A is a container, not a codec
A detail that explains a lot of confusion: M4A is a box, and the codec inside it is usually AAC. An MP4 video is the same box with a video track added. So "M4A versus AAC" is not a real choice, and a file called .m4a will play anywhere AAC plays, which in 2026 is everywhere.
AAC does the same job as MP3 and does it better at the same bitrate, which is why Apple, YouTube and most streaming services use it. MP3 survives because every device made since 1998 plays it, and because it is the format people have heard of. For a file that must open on something old or unknown, MP3. For anything else lossy, AAC in an M4A is the better default.
Ogg Opus is the technically best of the lot for speech and is what browsers use for calls, but it is still the one most likely to make a non-technical recipient ask what they are supposed to do with it.
Recording: capture lossless if you can
What a recorder produces is decided by the software, and a browser-based one is decided by the browser. The screen recorder here writes MP4 in Chrome and Safari and WebM in Firefox, because those are the containers the recording API offers; the audio inside is AAC or Opus respectively, and it is lossy from the first second.
Fine for a walkthrough or a meeting. For anything you might edit heavily afterwards, music, an interview that will be cut into pieces, a voiceover, a lossless capture is worth the space. A phone's voice memo app and most desktop recorders can be set to WAV or to Apple's lossless format, and the setting is worth finding once.
Editing: trim in WAV, then convert
Every edit on a lossy file that is then saved as lossy re-encodes the whole thing. Trim an MP3 in a tool that outputs MP3 and the untouched middle has been compressed twice. Do it three times across a project and the artefacts are audible on cymbals and sibilants.
The audio trimmer deliberately outputs WAV for this reason. It decodes whatever you give it, cuts it, and writes the result without any encoding step. The file is larger than the MP3 you started with, and that is the point: it has lost nothing. Make the MP3 at the end, once, from that.
A short fade at each cut, a few hundredths of a second, is the other thing an editor does that a plain cut does not. Without it the cut lands mid-wave and clicks. The trimmer applies one by default.
Sending: small, and in a format they can open
Email caps attachments somewhere between 10 and 25 MB, and messaging apps compress anything large. A ten minute WAV is 100 MB and will not go. A ten minute AAC at 128 kbps is about 10 MB and will.
For speech, 64 to 96 kbps AAC or MP3 is transparent to almost everyone. For music, 192 kbps and up. Higher bitrates than that on a lossy format are mostly reassurance.
If several files need to travel together, the create ZIP tool bundles them, though compressing an already-compressed MP3 saves almost nothing; the ZIP is for convenience, not size. For a video rather than audio, the same logic applies, and the image compressor is the equivalent for stills.
Keeping: the archive is the lossless one
Storage is cheap and getting cheaper; a lost original is lost forever. Keep the WAV, the FLAC or the lossless capture of anything that mattered to record, and treat every MP3 as a copy you can regenerate.
FLAC is the sensible archive format when WAV's size is a problem: it is lossless, about half the size of WAV, and opens in every modern player. The only reason to archive as WAV over FLAC is a tool in the chain that does not read FLAC, which is rarer every year.
Questions people ask
At the same bitrate, yes: M4A files normally contain AAC, which produces better sound than MP3 at equal size. MP3 wins only on compatibility with very old devices. For anything made in the last decade, an M4A plays everywhere an MP3 does.
No. The detail the MP3 discarded is gone, and the WAV simply stores the MP3's sound in a larger file. Convert to WAV only as a working format for editing, so that the edits themselves do not add another round of loss, then export back to a lossy format once.
64 to 96 kbps in AAC or MP3 is enough for speech to sound clean; 128 kbps is a safe default if stereo music plays under it. Mono halves the size again with no loss for a single voice.
Because trimming is an edit, and saving an edit in a lossy format re-encodes the whole file. WAV output means the cut loses nothing. The file is larger, and you convert it to MP3 or M4A afterwards, once, if you need it small.
Whatever the browser's recording API offers: MP4 with AAC audio in Chrome and Safari, WebM with Opus audio in Firefox. Both are lossy from the start, which is fine for walkthroughs and meetings, and a reason to use a dedicated recorder set to lossless for music or a voiceover you will edit.
FLAC, unless something in your workflow cannot read it. Lossless, around half the size of WAV, and supported by every current player and editor. The archive copy is the one you make lossy copies from, so it must be the one that has lost nothing.

