How to extract audio from a video file (MP3 / WAV)
Pull the audio track out of an MP4 or MOV and save it as MP3 / WAV. Uses ffmpeg.wasm in the browser so the source video never leaves the device.
When you only need the soundtrack
Record an internal workshop on Zoom or Google Meet and you get back a single MP4 or MKV file that packs the video and the audio together. If you hand 800 MB of footage to a transcriber so they can produce minutes, you are blowing past Slack’s 1 GB upload ceiling and Gmail’s 25 MB cap for almost nothing — the transcriber will only ever listen to the audio anyway. Pulling the same recording down to an MP3 typically drops a one-hour meeting to a few dozen megabytes.
The same operation shows up in lots of other workflows. You filmed a conference talk on your phone and want the speech for a podcast cut. You produced a YouTube video and need to extract a clean interview audio bed. You captured a rehearsal on a mirrorless camera and want to take the raw AAC track straight into the Music app on your iPhone to listen on the train. The common shape is “drop the video, keep the audio”, and that is exactly what the in-browser video-extract-audio is built for.
Pulling the audio out, in the browser
Open the video-to-audio extractor and drag your clips onto the page. It accepts the usual container formats — MP4, MOV, M4V, WebM, MKV — and you can drop several files at once. Once they load, the output format selector appears: pick mp3 for maximum compatibility, wav if you want a lossless raw PCM dump, m4a (AAC) for Apple ecosystems, or ogg (Vorbis) for an open-format target.
Press Extract audio and the in-browser ffmpeg.wasm build discards the video track and re-encodes only the audio. When it finishes, each file gets its own Download button, and a Download ZIP control bundles the lot. If the source happens to be a silent recording the tool stops with “This video has no audio track” rather than handing you an empty file. Keep the DevTools Network tab open during all of this and you will see exactly one set of requests during the first run — the WASM and script fetches — and nothing tied to the video content after that.
What ffmpeg.wasm is doing under the hood
The extraction is essentially the classic ffmpeg -i input.mp4 -vn -acodec <encoder> output.<ext> recipe. -vn (video none) throws away the video stream, and -acodec picks the encoder that matches the chosen container — libmp3lame for .mp3, pcm_s16le for .wav, aac for .m4a, libvorbis for .ogg. The only difference from a desktop FFmpeg run is that the native x86 binary is replaced by a WebAssembly build executing inside a Web Worker.
It helps to keep “container” and “codec” separate in your head. MP4, MOV, and WebM are just envelopes; what is actually inside can be AAC, Opus, Vorbis, or raw PCM. A naive approach would be to demux the AAC track straight into an .m4a with no re-encode, but compatibility across players and sample-rate mismatches make that surprisingly fragile in practice, so this tool always re-encodes to match the chosen extension. Pick wav (16-bit PCM) when you want lossless, and mp3 (libmp3lame) when file size matters more than the last few dB of fidelity. If you later want to retarget the result — say MP3 to FLAC, or M4A to OGG — feed the extracted file into audio-convert and you can bounce between mp3 / wav / m4a / ogg / flac without ever opening a desktop tool.
Why this differs from an upload-style converter
Search for “extract audio from video online” and you will hit dozens of sites that take your video upload and email you a download link to the resulting MP3. Internal meeting recordings tend to carry sensitive material — customer names, salary discussions, unannounced product plans — and pushing that through an unfamiliar vendor’s storage, even briefly, is a heavyweight decision under most companies’ information-handling rules. The terms of service usually include clauses like “we may retain logs to improve our service” or “uploaded content may be used to improve our models”, and no deletion request can verify what already touched the CDN caches and backup snapshots.
video-extract-audio has no code path that uploads the file at all. ffmpeg.wasm runs inside a Worker, the resulting audio Blob is exposed via an <a download> link in the same tab, and the source is on GitHub for you to audit. Instead of trusting a privacy policy after the fact, you can confirm with the Network tab and the public source that the recording never left the browser — which is the kind of guarantee you actually need when handling confidential footage.