How to extract audio from a video as mp3
- Step 1Open the audio track extractor — Load the audio-track-extractor tool. FFmpeg.wasm initialises in your browser the first time — after that it's cached. Nothing is sent to a server at any point.
- Step 2Drop your video file — Drag in an MP4, MOV, MKV, WebM, AVI, M4V, or TS file (one file at a time on this tool). On the Free tier the file can be up to 1 GB; Pro raises this to 10 GB and Pro + Media to 100 GB.
- Step 3Set Format to MP3 — The Audio output panel has a single Format dropdown. Choose MP3 for the 192 kbps re-encode. (WAV/FLAC give lossless masters; AAC stream-copies an AAC source untouched; Opus targets Discord voice — see the sibling guides.)
- Step 4Run the extraction — JAD runs
-i input -vn -c:a libmp3lame -b:a 192k out.mp3locally. MP3 always re-encodes (no source is natively MP3-in-MP4 in the common case), so there's a single encode pass — fast for speech, a little longer for long high-sample-rate music. - Step 5Save the file — On a Chromium browser the result can stream straight to a save location you pick; otherwise it downloads as
<name>.mp3. The output is the default (first) audio track of the source. - Step 6Verify and reuse — Play the MP3 to confirm. Need a lossless master instead? Use WAV/FLAC extraction. Cleaning a recorded podcast? See podcast audio extraction.
Format dropdown — what each option actually does
The audio output panel exposes exactly one control: Format. These are the five choices and the real FFmpeg behaviour behind each, as run by the in-browser engine.
| Format | FFmpeg codec args | Re-encode? | Best for |
|---|---|---|---|
| MP3 | -c:a libmp3lame -b:a 192k | Always (192 kbps) | Portable, universally playable speech + music |
| WAV | -c:a pcm_s16le | Always (16-bit PCM) | Editing/mastering, max compatibility uncompressed |
| AAC | -c:a copy | No — stream-copy | Bit-identical extract of an AAC source |
| FLAC | -c:a flac | Always (lossless compress) | Archive masters, smaller than WAV, lossless |
| Opus | -c:a libopus -b:a 128k | Always (128 kbps) | Discord voice messages, efficient speech |
Accepted inputs and tier limits
The extractor takes a single video file. Limits are file-size and batch-count only — there is no duration cap, so a 4-hour recording is fine if it fits under the byte limit.
| Aspect | Detail | Notes |
|---|---|---|
| Input containers | MP4, MOV, MKV, WebM, AVI, M4V, TS | The video stream is dropped with -vn |
| Files per run | 1 (no batch on this tool) | acceptsMultiple: false in the registry |
| Free tier | 1 GB / file | No minutes cap; multi-hour audio is fine |
| Pro tier | 10 GB / file, up to 5 files in other tools | Streams output to disk on Chromium |
| Pro + Media tier | 100 GB / file | For very long high-bitrate masters |
| Output | Audio file (MP3 here) | Type audio, named <source>.mp3 |
Cookbook
Real extract jobs and what the engine actually produces. Every job is one source file, one Format choice, one download.
Webinar MP4 to a shareable MP3
A 90-minute recorded webinar (MP4, ~700 MB) needs to become a podcast-style MP3 for the team to listen to on commutes. MP3 re-encodes at a fixed 192 kbps regardless of the source bitrate.
Source: webinar-2026-06.mp4 (90 min, AAC 256 kbps, ~700 MB) Format: MP3 Command: ffmpeg -i webinar-2026-06.mp4 -vn -c:a libmp3lame -b:a 192k webinar-2026-06.mp3 Result: webinar-2026-06.mp3 (90 min, MP3 192 kbps, ~123 MB) Video stream dropped; audio re-encoded to a portable file.
Phone clip soundtrack for a quick voice note
A short MOV from a phone (HEVC video + AAC audio) — you only want the speech. The video is irrelevant, so -vn removes it and you get a tiny MP3.
Source: IMG_4821.MOV (45 s, AAC audio, 88 MB with 4K video) Format: MP3 Result: IMG_4821.mp3 (45 s, 192 kbps, ~1.1 MB) The 4K video stream is discarded — file shrinks ~80x.
Music video MKV to MP3
An MKV with Vorbis or AAC audio. MP3 always re-encodes, so the chosen output is libmp3lame at 192 kbps no matter what the source codec was.
Source: liveset.mkv (Vorbis audio) Format: MP3 Command: ffmpeg -i liveset.mkv -vn -c:a libmp3lame -b:a 192k liveset.mp3 Result: liveset.mp3 (192 kbps CBR) For a lossless master instead, pick FLAC (see sibling guide).
Choosing MP3 vs AAC for a no-loss copy
If your goal is a perfect, untouched copy of the audio and the source is AAC, MP3 is the wrong choice — it re-encodes. Switch the Format dropdown to AAC for a stream-copy.
Goal: bit-identical copy of an AAC track Wrong: Format = MP3 -> re-encodes (lossy generation loss) Right: Format = AAC -> ffmpeg ... -vn -c:a copy out.aac See /video-tools/solutions/convert-video-soundtrack-to-aac-no-loss
Long lecture recording (3 hours) under the Free cap
There is no minutes limit — only the 1 GB Free file-size cap. A 3-hour 480p lecture in MP4 is usually well under 1 GB and extracts in one pass.
Source: lecture-week09.mp4 (3 h, 480p, ~820 MB) Format: MP3 Result: lecture-week09.mp3 (3 h, 192 kbps, ~248 MB) No duration cap hit; only the file-size limit applies.
Edge cases and what actually happens
Source has no audio track at all
FailsA screen recording or timelapse exported without sound has no audio stream. -vn removes video, leaving nothing for libmp3lame to encode, and FFmpeg exits non-zero — the tool reports an extraction error. There is no audio to recover; confirm the source actually contains sound before extracting.
MP3 always re-encodes — it is never stream-copy
By designThe MP3 path is hard-wired to libmp3lame -b:a 192k, so even if a source somehow already held an MP3 audio stream, choosing MP3 re-encodes it (a small quality loss). If you want a byte-perfect copy of the source audio, choose AAC (the only -c:a copy path) instead.
Bitrate is fixed at 192 kbps
By designThere is no bitrate slider — the only control is the Format dropdown. MP3 output is always 192 kbps CBR. If you need a higher-fidelity master, pick WAV (16-bit PCM) or FLAC (lossless); if you need a smaller speech file, Opus runs at 128 kbps.
Only the first audio track is extracted
By designThe engine extracts the default (first) audio stream — there is no track-picker UI on this tool, despite older copy that implied one. For a multi-language MKV where you need a non-default dub or a commentary track, the current build extracts the primary track only.
File exceeds the tier file-size cap
RejectedFree tier caps a single video at 1 GB. A large 4K master can exceed that. Upgrade (Pro = 10 GB, Pro + Media = 100 GB) or first trim the source losslessly with lossless-trimmer before extracting just the segment you need.
Very long high-sample-rate music in WASM
Slow but supportedFFmpeg.wasm runs in the browser, so encoding a multi-hour high-sample-rate source is CPU-bound and slower than native FFmpeg. It still completes — there is no minutes cap — but expect a longer single encode pass for big music files. Speech-rate recordings are quick.
Container has audio in a codec FFmpeg.wasm can't decode
FailsRare codecs (e.g. some proprietary or DTS variants depending on the build) may not decode in the WASM FFmpeg. The job exits with an error in the logs rather than producing silence. Transcode the source first or report the codec; most consumer MP4/MOV/MKV/WebM audio (AAC, Opus, Vorbis, MP3, FLAC) decodes fine.
Output plays but starts with a slight delay
PreservedIf the source audio track carries an initial timestamp offset (common in some camera and broadcast containers), the extracted MP3 preserves the original timing. Re-importing into an editor may reveal that delay; nudge the audio in your editor to re-sync if needed.
Frequently asked questions
Is my video uploaded anywhere?
No. The extractor runs entirely in your browser using FFmpeg.wasm. Your file is read locally, processed locally, and saved locally — nothing is sent to any server. This is what makes it safe for confidential meeting, interview, and medical recordings.
What bitrate is the MP3?
A fixed 192 kbps CBR via libmp3lame. There is no bitrate control on the page — the only setting is the Format dropdown. 192 kbps is transparent for speech and clean for general listening. For higher fidelity choose WAV or FLAC; for a smaller speech file choose Opus.
Which video formats can I extract from?
MP4, MOV, MKV, WebM, AVI, M4V, and TS. The tool drops the video stream with -vn and encodes whatever audio it finds, so a phone MOV, a downloaded WebM, and a broadcast TS all work.
Is the MP3 the same quality as the original audio?
MP3 is a re-encode, so it's a new lossy file at 192 kbps — close to the source for most listening but not bit-identical. If you need a perfect copy, choose AAC (stream-copy, only works when the source is AAC) or a lossless format (WAV/FLAC).
How long a video can I extract from?
There is no duration cap — limits are file-size only. The Free tier allows files up to 1 GB, so a multi-hour 480p or 720p recording usually fits comfortably. Pro raises this to 10 GB and Pro + Media to 100 GB.
Can I pick which audio track to extract from a multi-track file?
Not in the current build. The tool extracts the default (first) audio stream. There is no track-selection control on the page. For most files that is the main track; multi-language dubs or commentary tracks beyond the first aren't separately selectable yet.
Can I extract audio from several videos at once?
No — this tool processes one file per run (it doesn't accept multiple inputs). Run each video through individually. Batch processing exists on some other video tools, not the audio extractor.
Why did extraction fail with no output?
The most common cause is a source with no audio track (e.g. a silent screen recording or timelapse). -vn removes the video, and with no audio there's nothing to encode, so FFmpeg exits with an error. Confirm the source actually has sound.
Do I need to install FFmpeg or HandBrake?
No. The FFmpeg engine ships compiled to WebAssembly and runs inside the browser tab. There is nothing to install, no command line, and no system FFmpeg required.
What's the difference between this and just compressing the video?
This discards the video entirely and keeps only the audio. To shrink a video while keeping the picture, use a compressor like discord-compressor instead. The audio extractor's whole job is to give you an audio-only file.
Will the MP3 keep chapters or metadata?
MP3 output is a plain audio re-encode; it won't carry over video chapters. If you need the source's chapter markers separately, use chapter-extractor. To strip identifying metadata from the original video first, see metadata-scrubber.
I want the audio for a podcast — is there a better path?
Yes — for video-recorded podcasts, the extract podcast audio guide covers feeding the result into mastering. For a lossless editing master, use WAV/FLAC extraction.
Privacy first
Every JAD Video tool runs entirely in your browser via WebCodecs and FFmpeg (WebAssembly). Your video files never leave your device — verified by zero outbound network requests during processing.