Skip to main content

Extract audio from video online

Extract audio tracks from video files with full control over format, quality, and processing.

Free & unlimited
All processing happens in your browser — files never leave your device. The audio engine loads once from a CDN.

Take the soundtrack out of a video file and save it as audio. The video is opened by ffmpeg running in your browser, the video stream is dropped, and only the audio is encoded into the format you choose — MP3, WAV, AAC, FLAC, OGG Vorbis or Opus. Because the video stream is discarded rather than converted, this is much faster than a full video conversion and the output size depends only on the audio settings. You control bitrate from 64 to 320 kbps for the lossy formats, sample rate from 22.05 kHz up to 96 kHz or matched to the source, and mono or stereo. Seven container formats are accepted, covering everything a phone, a camera or a screen recorder is likely to produce.

Key facts about Extract audio from video online

Key facts about Extract audio from video online
Mechanismffmpeg runs with -vn, which drops the video stream entirely rather than transcoding it
Accepted containersMP4, WebM, MOV, AVI, MKV, FLV and WMV, up to 200 MB
Output formatsMP3 (libmp3lame), WAV (PCM), AAC, FLAC, OGG Vorbis and Opus (libopus)
Bitrates64, 128, 192, 256 and 320 kbps for the lossy formats
Sample ratesMatch the source, or force 22.05 kHz, 44.1 kHz, 48 kHz or 96 kHz
ChannelsAuto, mono or stereo — downmixing to mono halves the size of a talking-head recording at no real cost
Typical source rateVideo audio is usually 48 kHz AAC; leaving sample rate on auto avoids an unnecessary resample
Lossless optionExtracting to WAV or FLAC keeps everything the source audio had, but does not undo the video's original lossy encode
What is lostVideo, subtitles, chapters and every audio track beyond the first
Size cap200 MB, set by the WebAssembly heap rather than by policy

What happens to your file

The video is processed by ffmpeg compiled to WebAssembly, running in a Web Worker inside this tab. Your file is written into that worker's in-memory filesystem, the audio stream is extracted and encoded, and the result is read back as a blob — nothing is uploaded and no copy is kept anywhere. The one network request involved is the first-time download of the ffmpeg core, about 32 MB of fixed public code from the jsDelivr CDN, which your browser caches and reuses across every audio and video tool here. That request contains no part of your video.

About this tool

  1. 1

    Add the video

    Drop an MP4, WebM, MOV, AVI, MKV, FLV or WMV up to 200 MB. Duration and size are read from the file header.

  2. 2

    Choose the output format

    MP3 for maximum compatibility, AAC if the source was MP4 and you want to stay in the same codec family, WAV or FLAC when the audio will be edited afterwards, Opus for the best size-to-quality ratio.

  3. 3

    Set the bitrate

    192 kbps is a safe default for music, 128 kbps is plenty for speech, 64 kbps only makes sense with Opus. Lossless formats ignore this control.

  4. 4

    Leave sample rate on auto

    Video audio is almost always 48 kHz. Forcing 44.1 kHz adds a resample for no benefit unless the destination specifically needs CD rate.

  5. 5

    Pick channels

    Auto keeps the source layout. Mono halves the file size and is the right choice for an interview, a lecture or any single-speaker recording.

  6. 6

    Extract and download

    Progress is shown as a percentage while ffmpeg runs, and the finished audio can be played back in the page before you save it.

Specs & compatibility
Input containersMP4, WebM, MOV, AVI, MKV, FLV, WMV
Maximum file size200 MB
Output formatsMP3, WAV, AAC, FLAC, OGG, Opus
EngineSingle-threaded ffmpeg.wasm 0.12.10, GPL build
Engine download~32 MB, fetched once and cached by the browser
SpeedFaster than a video conversion because the video stream is never decoded; typically a fraction of the clip's duration
BrowsersChrome, Edge, Firefox and Safari 15+
Multiple audio tracksOnly the first audio stream is extracted; alternate language tracks are not selectable
  • Extracting to WAV does not improve anything. The audio inside a video is already lossy; WAV just stores that same audio uncompressed and much larger.
  • Keep the sample rate on auto. Video soundtracks are 48 kHz by convention, and resampling to 44.1 kHz is a small quality cost with no upside.
  • Downmix to mono for anything with one speaker. It halves the file with no audible loss, because the two channels are near-identical anyway.
  • If you only want a section, extract the whole soundtrack and then cut it with the audio trimmer, which can do that without re-encoding.
  • Opus at 96 kbps beats MP3 at 160 kbps for speech, which matters if you are archiving a lot of lecture recordings.
  • 200 MB is the ceiling, and video files reach it quickly. If your clip is larger, trim the video first or use a lower-resolution export of it.
  • For screen recordings with quiet narration, follow up with the audio normalizer rather than raising the bitrate.
  • Seven input containers including MP4, MKV, MOV and WebM
  • Six output formats from MP3 to Opus
  • Bitrate control from 64 to 320 kbps
  • Sample rate matched to source or forced to a standard rate
  • Mono or stereo output
  • Video stream dropped rather than transcoded, so extraction is fast
  • In-page playback of the extracted audio before download
  • Saving the audio from a recorded lecture so you can listen on a phone
  • Getting a podcast-ready MP3 out of a video interview
  • Pulling the soundtrack off a screen recording to edit the narration separately
  • Extracting music from a concert video you filmed yourself
  • Turning a long meeting recording into a much smaller audio-only file for archiving
  • Producing a WAV from a video clip to drop into an audio edit
It depends on the format you choose. Extracting to WAV or FLAC preserves the audio stream's decoded content exactly, so nothing extra is lost — though the audio inside the video was already lossy, and that damage is permanent. Extracting to MP3, AAC, OGG or Opus means decoding the original lossy audio and re-encoding it, which costs one further generation. At 192 kbps or above that is usually inaudible, but for material you will edit further, extract to WAV and encode once at the end.
MP3 if the file needs to play anywhere, including old hardware. AAC if the source was an MP4, since its audio is almost certainly AAC already and staying in the family is the closest thing to a like-for-like copy. Opus if you want the smallest file for a given quality, particularly for speech. WAV or FLAC when the audio is going into an editor. FLAC is the sensible archive choice: lossless, but roughly half the size of the equivalent WAV.
The limit is 200 MB, and that is a memory constraint rather than a policy. ffmpeg.wasm holds the input file inside a single WebAssembly heap of roughly 2 GB alongside its working buffers, and video files are large by nature — a few minutes of 1080p can pass 200 MB easily. If your file is over the cap, either trim the video first with a video editor, export a lower-bitrate version, or use a screen recorder setting that produces a smaller file to begin with.
No. The tool extracts the first audio stream in the container, which is what you want in virtually every real case. Files with several audio tracks — a Blu-ray rip with multiple language dubs, or a video export with a separate commentary track — will always give you the first one. Selecting a specific stream requires a track picker that this tool does not have. For multi-track work, a desktop tool such as native ffmpeg or a video editor is the right choice.
Yes, substantially. ffmpeg is invoked with -vn, which tells it to ignore the video stream entirely rather than decoding and re-encoding it. Since video decoding and encoding is where nearly all the work in a video conversion happens, dropping it means only the audio — a tiny fraction of the data — passes through the encoder. A ten-minute clip that would take several minutes to transcode typically extracts in a fraction of that, even on single-threaded WebAssembly.
No. It is processed by ffmpeg compiled to WebAssembly, running in a Web Worker on your own machine. The file goes into that worker's in-memory filesystem, the audio comes back out as a blob, and nothing crosses the network. The only request the tool makes is fetching the ffmpeg core itself — about 32 MB of fixed public code from the jsDelivr CDN, cached by your browser and shared with every other media tool here. Your video plays no part in that download.

Related tools

View all

Part of Audio tools that run in your browser

Updated

We use anonymous analytics to improve ToolChamp. No personal data is stored or sold. Privacy Policy