Video · 16
Video to MP3
Pull the soundtrack out of a screen recording, a phone clip, or a timeline export. You get an MP3 and the picture stays behind.
How to extract a clean MP3
Use the original export, not a video that has already been crushed for chat. The encoder can only keep what is in the file. A 192 kbps MP3 made from a tiny social upload will sound like that upload.
Pick the bitrate for the destination, not for the ego of the file size. Speech for a podcast host is fine at 128 kbps. A music bed you will upload to YouTube is happier at 192 kbps. Choose 320 kbps when someone else will master or archive the file and you do not want another generation of loss on top of the camera audio.
The command is a straight extract: no video stream, LAME at a constant bitrate. There is no loudness pass and no noise reduction. If you need the vocal lifted out of a song, use the vocal remover on the MP3 or, better, on a WAV you still have.
Files over 250 MB are refused. A long 4K master is a poor way to move audio. Export an audio mixdown from the editor when the picture is huge and you only wanted the sound.
Tips
- Phone videos in MP4 or MOV are the usual case and work.
- Keep the original video. This tool does not replace it.
- If you will separate vocals next, 192 kbps or higher gives the model more to work with than 128.
- Play the preview before you send the file. A wrong audio track in a multi-track export shows up immediately.
FAQ
Does this improve a bad recording?
No. It copies the audio track that is already in the file and encodes it as MP3 at the bitrate you pick. Noise, clipping, and room tone stay.
Which bitrate should I use?
128 kbps is enough for speech. 192 kbps is a sensible default for music you will put under a YouTube video. 320 kbps is for when the destination should stay close to the source and file size is secondary.
Where does ffmpeg run?
In a single-threaded WebAssembly build inside the tab. The first run downloads about 31 MB and caches it. Later jobs reuse that copy. This build does not need SharedArrayBuffer by itself; the site still sends isolation headers because the AI tools do.
What if the video has no audio?
ffmpeg will fail and the page shows the encoder message. Screen recordings muted at export are the usual cause.
Is the video uploaded to extract the audio?
No. The file is read by the browser and handed to the WASM module in memory. Nothing is posted to a server.
License
The JavaScript wrapper from ffmpeg.wasm is MIT. The core binary shipped here is a GPL build of FFmpeg because it includes libmp3lame for the MP3 encoder. Source for FFmpeg is at ffmpeg.org. Using the extractor means you are running that GPL binary in your browser.