Get the audio out of a video file as a WAV. Drop the video in, optionally trim it or normalise the level, and download. Your browser decodes the audio track and ignores the video, so nothing is uploaded and there is no conversion queue. One thing worth being plain about, because most tools in this category are not: this decodes the compressed track and writes uncompressed PCM. It does not lift the original AAC or Opus stream out untouched — that needs a demuxer this does not ship. Nothing is lost a second time, but the file is much larger than the video's audio was.
| Aspect | Result |
|---|---|
| Quality | No second compression — but decoded, not copied |
| Size | ~10 MB per minute, stereo 44.1kHz 16-bit |
| Channels | Whatever the source had |
| Format | WAV only — convert afterwards for MP3 |
Not in the strict sense, and it is worth knowing the difference. A true extraction copies the encoded AAC or Opus stream into a new container without touching it. This decodes that stream and writes uncompressed PCM. Nothing is compressed a second time, so no quality is lost beyond what the video already lost — but the output is not byte-identical to the embedded track, and it is much larger.
Because WAV is uncompressed. A minute of stereo at 44.1kHz is around 10MB no matter what bitrate the video's audio used. Drop to 16-bit if you have it higher, or compress the result afterwards if size matters.
Whatever your browser can decode the audio of — typically MP4, MOV, WebM and M4A. If a file is refused, your browser has no decoder for that audio codec; there is no conversion step here that could work around it.
Not directly. Encoding MP3 in a browser needs a bundled encoder of several hundred kilobytes, and re-compressing audio that was already compressed loses quality again. Take the WAV and convert it if you specifically need MP3.
Yes — however many channels the track has are decoded and written, interleaved correctly. A stereo source comes out stereo.
Yes. Set a start and end before exporting, which saves running it through a separate trimmer. Peak normalisation is available too, though that lifts the loudest sample rather than making things sound equally loud.
No. It is decoded in your browser, which matters for footage you would not want to hand to a website.
The video file is handed to `decodeAudioData`, which decodes its audio track and ignores the video entirely — no demuxer, no WASM, no upload. That gives raw float samples, which are then written out as a RIFF/WAVE file at 16, 24 or 32-bit float. Being plain about what that means matters: the result is a decode of a compressed track, not the original track lifted out untouched. A true extraction would copy the encoded AAC or Opus stream into a new container without re-encoding, which needs a demuxer this deliberately does not ship; what happens here is decode to PCM and write PCM. The audio is not re-compressed, so nothing is lost a second time, but the file is much larger than the source — a minute of stereo at 44.1kHz is about 10MB whatever the video's audio bitrate was. Two details in the writing. Samples are interleaved by frame rather than by channel, because WAV stores them L R L R and writing one whole channel and then the next produces a file that plays as two halves. And 16-bit conversion scales negative and positive by their own limits, since signed 16-bit runs from -32768 to +32767 and treating both directions as 32768 overflows a full-scale sample into a loud click. The decoder runs at the file's own sample rate where that is knowable and at the browser's audio rate otherwise, and the rate actually used is displayed rather than implied. Optional trimming and peak normalisation share the same engine the audio trimmer uses. What this cannot do is read a container your browser has no audio decoder for, which is stated up front rather than discovered after a wait.