Convert audio between formats, without uploading it
Drop in as many files as you like, choose an output format, and press Convert. The tool converts between MP3, M4A, WAV, FLAC, OGG, Opus, AAC and sixteen other formats, and it will pull the soundtrack out of a video file just as happily. It is free, needs no account, adds no watermark, and has no file size limit beyond what your own machine can hold.
Nothing is uploaded. A real build of FFmpeg, compiled to WebAssembly, runs inside the tab and reads your files straight off your disk. You can confirm it by loading the page, disconnecting from the network, and converting anyway — it works offline.
Headerless PCM, and how to find out what it is
A .pcm or .raw file is nothing but samples. It does not record how fast it was captured, how wide each sample is, or how many channels it interleaves — so the same bytes are a voice recording at 8000 Hz mono and a burst of noise at 48000 Hz stereo, and nothing in the file says which. Most converters either refuse these files or guess silently and hand you static.
Press PCM on the file's row. If you know the recording settings, set them. If you do not, turn on Try-Yourself-Mode and step through the combinations — nine sample rates, four bit depths and five channel counts, 180 in all — listening for the one that stops sounding like noise. Untick anything you can rule out and the search gets shorter. Each step plays instantly, and the playhead is positioned by percentage of the file rather than by time, so every combination is judged on exactly the same bytes.
Everything defaults to leaving your file alone
Sample rate and channel count both start on same as original, and that means the setting is genuinely not applied rather than applied with a value the tool picked. A 5.1 mix stays 5.1; a 96 kHz master is not quietly resampled to 44.1.
Better still, where the output format can carry your existing track as it is, it is copied rather than re-encoded — so pulling AAC audio out of an MP4 into an M4A costs nothing and loses nothing at all. The row tells you which happened. Setting a quality, sample rate or channel count makes a copy impossible by definition, so any of those re-encodes the track.
Output formats
| Group | Formats | Best for |
|---|---|---|
| Most common | MP3, M4A, WAV, FLAC, OGG, Opus, AAC | Everyday conversion. MP3 plays on essentially anything; Opus gives the best quality per kilobyte of anything here. |
| Lossless | FLAC, ALAC, WAV, AIFF, WavPack, TTA, CAF | Archiving and mastering, where every sample has to survive. FLAC is about half the size of WAV and loses nothing. |
| Broadcast and cinema | AC-3, E-AC-3, DTS, TrueHD, MLP | Surround delivery, DVD and Blu-ray authoring. |
| Legacy and other | WMA, MP2, AU, MKA, WebM, MP4 | Older players, broadcast chains and hardware that will not take anything newer. |
Questions
Are my files uploaded?
No. There is no server to upload to. The tool downloads a copy of FFmpeg the first time you convert something, and from then on all the work happens in your browser. Your files never leave the machine.
Can it take the audio out of a video?
Yes — drop the video in and pick an audio format. Only the soundtrack is kept. If the format you choose can carry that soundtrack untouched, it is extracted rather than re-encoded, which takes about as long as copying the file.
How fast is it?
Fast enough that it is rarely the thing you are waiting for. A minute of stereo audio converts in well under a second for most formats — around 0.6 s to 320 kb/s MP3, 0.1 s to FLAC, 2 s to Opus. The slow part is the one-off download of FFmpeg itself, about 31 MB, which happens once per session.
Why is there no AMR?
Because the WebAssembly build of FFmpeg this uses has no AMR encoder compiled into it at all, and there is no native substitute — offering a button that cannot finish is worse than not offering it. Every other format the desktop version of this tool supported is here.
Which quality setting should I use?
For MP3, the default of roughly 320 kb/s is more than the format needs to sound transparent, which is the right property for a default. If you would rather have a smaller file at the same perceived quality, VBR V0 in Advanced settings is the usual answer. For anything lossless — FLAC, ALAC, WAV, WavPack — there is no quality setting, because every sample is kept either way.
Does it work on a phone?
Yes, better than the video tools do. Audio encoding is light, so a phone handles it comfortably; the main cost is the initial FFmpeg download over mobile data.