Extraire audio d'une vidéo en MP3: browser guide

Extraire audio d'une vidéo en MP3: the short answer

To extract audio from a video as MP3, you pull the audio stream out of the video container and save it as a standalone .mp3 file — either by copying the stream untouched when it is already MP3, or by re-encoding it with an MP3 encoder when it is not. You can do that in a browser, with a desktop converter, or through an online service that uploads your file.

Two words are worth pinning down. A container (MP4, MOV, WebM, MKV) is the wrapper holding the video stream, one or more audio streams and the metadata; a codec (H.264, AAC, MP3, Opus) is the compression format used inside that wrapper. Opening the wrapper is demuxing — the FFmpeg documentation calls demuxers "splitters, because their main function is splitting a file into elementary streams". Turning AAC or Opus audio into MP3 is transcoding, and it costs a little quality because MP3 is lossy.

The real question is not which app to open, but whether you need to re-encode at all. If the audio track inside the video is already MP3, a stream copy is instant and lossless. If it is AAC — the usual case in MP4 and MOV files, and in everything an iPhone records — you must re-encode, and the only thing left to control is the bitrate.

How to extract the audio in your browser

This is the route we built and test ourselves: the Free Audio To MP3 Converter runs inside the page, so your video is never handed to a server. That matters when the clip is a work meeting, a family recording or anything you would rather keep to yourself. It also removes the upload wait on a slow connection, and there is no queue behind other people's jobs.

1. Play the video first. If your player shows no picture or no sound, the file is damaged or uses a codec the browser cannot decode, and a converter will fail for the same reason.
2. Drop the file on the page, or pick it with the file button. MP4, MOV, WebM, M4A, WAV and OGG files are accepted.
3. Choose the output bitrate: 128 kbit/s as a default, 192 kbit/s for music, 96 kbit/s for speech-only recordings. Pick a high bitrate only when the source is worth it.
4. Run the conversion and wait for it to finish. Everything is computed locally, so a long video takes roughly as long as your device needs to decode it.
5. Download the .mp3 and play the first and last ten seconds before you delete the original video.

What the browser is actually doing

Browsers decode media natively. MDN's Web Audio API reference notes that an AudioBuffer can be "created from an audio file using the BaseAudioContext.decodeAudioData method", and that a page can take an audio or video element as its source. So the page decodes your video's audio into memory as raw samples, then an MP3 encoder packs those samples back into frames. That memory step is why a huge camcorder file can stall a phone: the samples sit in RAM while the job runs. Convert large files on a desktop when you can.

Which bitrate to pick

libavcodec's codec documentation sets the default audio bitrate at 128 kbit/s when nothing else is specified (option ab, encoding, audio), and that is a sound starting point for a video soundtrack. A higher number does not recover detail the original recording never carried. If the source track is AAC at 128 kbit/s, exporting MP3 at 320 kbit/s mostly gives you a bigger file with the same information in it.

Desktop routes: FFmpeg and VLC

FFmpeg is the engine inside a lot of converters and it is free to install. Its manual describes it as "a universal media converter" that reads an arbitrary number of inputs, filters and transcodes them, and writes any number of outputs. One command extracts the audio track from a video:

ffmpeg -i input.mp4 -vn -c:a libmp3lame -b:a 192k output.mp3. Those flags are the standard FFmpeg ones: -vn drops the video stream, -c:a libmp3lame selects the MP3 encoder, and -b:a 192k sets the audio bitrate. Swap the encoder for -c:a copy and you get a lossless, near-instant export, but only when the existing track is already MP3. VLC reaches the same result through Convert / Save in the Media menu, which is friendlier and has fewer knobs. The trade-offs are real: you install software, you click through a wizard or type a command, and on a locked-down work laptop you may not be allowed to install anything at all.

Browser, desktop or upload service?

None of the three is always right. A browser tool wins on privacy and speed to first use, and loses on very large files and on batch work. A desktop tool wins on control and volume, and loses on setup time. An upload service wins when you have no computer at all, and loses on confidentiality and on what happens to the copy sitting on someone else's disk.

Common mistakes and files that fail

Mistake number one is extracting the wrong audio track. A film file or a downloaded video often carries several audio streams, such as an English dub, a French dub or a director's commentary. On the command line, ffmpeg -i input.mp4 lists them before you convert, so you can choose one with -map. In a browser you get whichever track the decoder picks first. Spot it by listening, not by looking at the file size.

The second mistake is converting twice. If you already extracted an MP3 and then re-encode it to a different bitrate, the audio is decoded and compressed a second time and loses more detail. Keep the first MP3, or go back to the original video instead. You can hear the problem before you can measure it: cymbals and the letter s smear first.

Some files fail outright. DRM-protected downloads and offline copies from subscription services are encrypted, so no browser decoder can read them and the conversion errors out rather than producing silence. A screen recording with no audio track gives you nothing to extract. Very long files can exhaust memory on a phone. And when a service asks you to paste a video URL, you are no longer extracting audio locally. You are asking that service to fetch and process someone else's stream, which is a different legal question from converting your own recording.

Quality, privacy and the limits of each route

MP3 is lossy perceptual coding: Wikipedia's MP3 article describes the algorithm taking advantage of auditory masking, discarding sound the ear is unlikely to notice and reconstructing the rest. That is why an MP3 will never be bit-identical to the audio inside the video. When you need a faithful copy — archiving, audio forensics, or another round of editing — export WAV or FLAC and keep MP3 for playback.

Privacy is the part we can speak to from our own data. Our MP4 to MP3 page has taken 18 sessions in the current rolling window and the free video converter 9, according to our internal analytics. The design rule behind both is that files are processed in the browser and never uploaded. What those counters do not tell us is which files people convert: we count sessions, not contents, and we do not inspect them. That is deliberate, and it is also a limit of what we can claim.

Before you use the MP3 in public, check your rights. Extracting audio from a recording you made is one thing; taking the soundtrack of a commercial film, a paid course or a video you do not own and republishing it is copyright territory. This is general information, not legal advice, so check with a qualified person before relying on it for commercial use. For more on that boundary, see our notes on converting a video to MP3 without uploading and on choosing between online video-to-audio converters, where we list what each route costs you.

No converter repairs a bad source, ours included: if the original audio is muffled, mono or clipped, the MP3 inherits all of it. As of 2026-10, the honest summary is that extraction is easy, quality is decided before you convert, and the safest default for private footage is the route that never uploads the file.