How to extract audio from a video

A recorded lecture you want to hear on a run. A song someone filmed at a concert. A client call that was captured as a screen recording. An interview that has to go to a transcriber. In all of these the picture is dead weight: it takes up 90 to 99% of the file, and the part you actually need is the audio track.

This guide walks through the whole job: getting the sound out, choosing a bitrate that matches the source, cutting the piece you need, and fitting the result into an email, a messenger or an upload form.

What you are actually extracting

A video file is a container with at least two streams inside: video and audio. On phones the audio is almost always AAC, typically at 96 to 256 kbps. Screen recorders and browsers often produce WebM with Opus audio. Extracting means dropping the video stream and saving only the sound.

One thing extraction cannot do is improve the sound. The audio inside the video is already compressed, and re-encoding it at a higher bitrate does not bring back anything the phone's encoder removed.

The quick way: MP4 to MP3 in the browser

1. Open MP4 to MP3 and drop the file. MP4, MOV (including iPhone recordings) and WebM are accepted, up to 2 GB per file.
2. Pick a bitrate: 128, 192, 256 or 320 kbps. The next section explains which.
3. Wait for the conversion and download the MP3. Several files can be dropped at once; they are processed one after another with the same bitrate.

The conversion runs with FFmpeg compiled to WebAssembly, so the video is read from your disk and never uploaded. The engine is about 30 MB and is downloaded once on first use, then cached by the browser.

If the clip has several audio tracks, for example two languages, one is picked automatically and the others are ignored. A 5.1 track is folded down to stereo, because MP3 has no surround mode. A video recorded without sound has nothing to extract, and the conversion fails.

Choosing the bitrate

At a constant bitrate the file size follows directly: kbps divided by 8 is kilobytes per second.

• 128 kbps, about 0.96 MB per minute. Speech of any kind: lectures, meetings, calls, voice notes, podcasts recorded on a phone.
• 192 kbps, about 1.44 MB per minute. The default. A safe choice for mixed content and most music.
• 256 kbps, about 1.9 MB per minute. Music from a good source that you will listen to on decent headphones.
• 320 kbps, about 2.4 MB per minute. The MP3 maximum. Useful only when the source itself is high quality.

Match the bitrate to the source, not to an ideal. A phone video with 128 kbps AAC audio does not sound better as a 320 kbps MP3; it just gets 2.5 times bigger. For a recording you will edit further, 256 or 320 kbps leaves headroom so the next encode does not add audible artefacts.

An hour-long lecture is about 58 MB at 128 kbps and 86 MB at 192 kbps.

Built-in ways that are good enough

Sometimes you do not need a converter at all.

macOS

QuickTime Player can do it: open the video, choose File, then Export As, then Audio Only. The result is an M4A file with AAC audio, which plays in Music, on iPhones and in most modern players. If the destination insists on MP3, use the browser tool instead.

iPhone and iPad

There is no one-tap button in Photos. The Shortcuts app has an Encode Media action with an audio-only option, which works but takes a few minutes to set up. For a one-off job, opening this page in Safari and picking the video from Photos is faster. Phones decode slowly, so for long videos a computer is the better place.

Windows and Android

Windows has no built-in way to save only the audio from a video in one step, and on Android it depends on the phone maker's gallery app. The browser tool works the same everywhere.

Keep only the part you need

A two-hour recording rarely needs to be shared whole. Open the MP3 in MP3 cutter: it draws a waveform, so pauses between sentences are easy to spot, and two sliders with a 0.1-second step set the start and end.

Tips for clean cuts:

• Place both points in silence, in the flat gaps of the waveform, so you do not chop a syllable.
• Leave 0.1 to 0.3 seconds of margin before the first word.
• To pull several quotes from one file, trim, download, move the sliders and trim again. The file stays loaded.

The cut is re-encoded as 192 kbps MP3, which lands exactly where you put the sliders instead of snapping to the nearest MP3 frame. The same tool makes ringtones: 20 to 30 seconds starting on the hook is a sensible length.

Fitting size limits

Common destinations and how to get under their limits:

• Email attachments. Many mail services cap attachments at around 20 to 25 MB. At 128 kbps that is roughly 20 to 25 minutes of audio. For longer recordings, use mono and a lower bitrate.
• Upload forms for courses, job applications and support tickets often have limits of 10 MB or less.

For speech, Compress MP3 can take the file down to 64 or 96 kbps and mix it to mono. A voice survives this very well: a one-hour talk at 64 kbps mono is about 29 MB, compared with 86 MB at 192 kbps stereo. Music does not take mono and 64 kbps gracefully; keep it at 128 kbps stereo or more.

When MP3 is not the right output

MP3 plays everywhere, which is why it is the default. Other formats make sense in specific cases:

• M4A is native to Apple devices and sounds slightly better than MP3 at the same bitrate.
• WAV is what some audio editors and broadcast workflows insist on. It is uncompressed: about 10 MB per minute of stereo.
• FLAC is lossless and smaller than WAV, useful for archiving.

The Audio converter writes all of these, but it takes audio files rather than video. Extract the soundtrack first, at 320 kbps if quality matters, then convert the MP3. Since the audio in a phone video was lossy to begin with, a WAV made this way does not lose anything you would hear; it simply satisfies software that will not accept MP3.

Need the words, not the sound

If the goal is a transcript, extract the audio first and send the MP3 to transcription. Uploading a 14 MB MP3 instead of a 1 GB video is much faster and gives the same text. Transcription runs on a server; without an account it handles recordings up to 10 minutes once a day, so long lectures need to be cut into parts with the MP3 cutter, ideally at pauses.

Troubleshooting

• The conversion fails immediately. Check that the video actually has sound: play it and watch the volume. Screen recordings are often made with audio disabled.
• The file is rejected. Only MP4, MOV and WebM are accepted. An MKV or AVI file can be converted with MKV to MP4 or AVI to MP4 first.
• A large file stalls on a phone. The whole video has to fit in the tab's memory. Try a desktop browser, or the server option offered on the page: one free run a day for files up to 500 MB, or up to 4 GB with Premium. Server files are deleted automatically within about an hour.
• The cover image or thumbnail is missing. Picture streams are dropped along with the video.

Checklist

1. Confirm the video has an audio track and is MP4, MOV or WebM.
2. Choose 128 kbps for speech, 192 kbps for mixed content, 256 to 320 kbps only for good music sources.
3. Extract with MP4 to MP3 and check the result plays from start to end.
4. Cut what you do not need with the MP3 cutter, at pauses.
5. For upload limits, compress speech to 64 or 96 kbps mono.
6. Convert to WAV or FLAC only if the destination requires it.