Convert an audio file
Read almost any audio or video file and write the format you need.
How it works
Five output formats cover nearly every reason anyone converts audio, and they divide cleanly by what they are for. Two keep every sample. Three throw samples away in exchange for size. Picking correctly is mostly a matter of knowing which of those two things the situation calls for.
The lossless pair
WAV stores raw measurements with no compression at all, which makes it universally readable and enormous: about 10.1 MB per minute in stereo at CD settings. It is what audio editors, samplers, phonetics software and speech pipelines ask for, and it is the wrong thing to keep a library in.
FLAC stores the same samples exactly, using prediction and entropy coding to reach roughly half the size. Decoding gives back the original bit for bit. Use it for archiving a recording or a rip, and check the destination reads it, since Apple devices generally do not.
The three lossy formats, and when each wins
MP3 is the oldest and the least efficient, and it plays on everything ever built. Choose it when the destination is a car stereo, a gym machine, an older portable player, or anything whose specification you cannot look up. Compatibility is the entire argument and it is usually the decisive one.
M4A carries AAC, which was designed to replace MP3 and does so on the numbers: at a given bitrate it holds up better, particularly below 128 kbps. Everything made since roughly 2012 plays it. Choose it when the destination is a phone, a modern car, a website or a video editor.
OGG carries Opus, which beats both at low bitrates by a wide margin and is the codec behind most voice chat and much web streaming. Speech at 32 kbps in Opus is startling against what MP3 does with the same budget. Choose it for the web, for voice, or wherever you control what plays the file.
The rule that saves the most regret
Convert from the best copy you have, once. Every lossy encode discards detail permanently, so a chain of them compounds: the second encoder cannot tell music from the first encoder's artefacts and spends bits on both. Going from lossless to lossy is a single clean decision. Going from lossy to lossy is a compromise worth making deliberately, at a bitrate at or above the source's own.
Video files work as input too. The picture is discarded and the soundtrack converted, so an MP4 recording of a meeting becomes an audio file of the meeting.
Questions
What can I put in?
MP3, WAV, M4A, AAC, FLAC, OGG, Opus, Vorbis and AIFF-style PCM, plus the audio track from MP4, MOV, WEBM, MKV, AVI and MPEG transport streams. What the file is named matters less than what is inside it, because the container is parsed rather than trusted. Dolby AC-3 and E-AC-3 are the notable gap: browsers ship no decoder for them.
Is anything uploaded?
No. The file is read by the page, decoded by the browser's own audio decoders, and re-encoded in a Web Worker on your machine. Open developer tools, watch the Network tab, and convert something: after the page and the relevant encoder module load, there are no further requests. This is verifiable in about ten seconds, which is the only kind of privacy claim worth making.
Which format should I pick if I have no idea?
MP3 at 192 kbps, because it plays everywhere and that bitrate is above the point where most people can hear the difference on most equipment. If you know the file is going onto a phone or a website, M4A at 128 kbps gives you a smaller file at the same perceived quality. If it is going into an audio editor, WAV.
Why is there no batch mode?
Each conversion holds the decoded audio in memory, and running several at once is how a phone browser runs out of it. There is no limit on how many files you convert in a row and nothing is counted, so a folder is a sequence of drops rather than a queue. The conversions themselves take a second or two each.