Pull the sound out of an MP4, MOV, WebM or MKV video and save it as MP3. Pick the bitrate.
This tool extracts the audio from a video by asking your browser to decode the file's sound and then compressing that sound to MP3 on your device. The browser's Web Audio decoder reads the video container, finds the audio track and returns raw samples. The picture is ignored and never decoded. The samples then go to LAME, the MP3 encoder, compiled to WebAssembly and run in a background worker so the page stays responsive. The result is a new MP3 file with the video's name and the .mp3 ending. Nothing is cut from the sound: the whole audio track is converted, from the first second to the last.
Whether a video opens depends on your browser and on the audio codec inside it, not on the file extension. An MP4 or MOV file with AAC audio, which is what phones and most screen recorders produce, is the most widely readable. WebM files from a browser recorder carry Opus or Vorbis audio, which Chrome, Edge and Firefox decode. MKV, AVI, WMV, FLV and old 3GP files are less certain, and audio in codecs such as AC-3 or DTS may not play at all. When the browser cannot decode the sound, the tool says so in a fixed message instead of producing an empty file. A video with no audio track fails the same way. A video protected by copy protection cannot be read. If one browser refuses a file, another may open it.
At a constant bitrate every second of sound takes the same space: seconds times kilobits per second times 1000, divided by 8. A 3-minute clip is 180 seconds, and at 192 kbps that is 180 times 192,000 divided by 8, or about 4,320,000 bytes. That is 4.32 MB counting a million bytes to the MB, and 4.1 MB in the size lines of this tool, which count 1 MB as 1024 KB. At 128 kbps the same clip is about 2,880,000 bytes and at 320 kbps about 7,200,000. These are estimates of the audio data; the real file is a little larger because of frame headers and padding. The audio in the video was already compressed, usually as AAC, so a higher bitrate does not restore detail that was lost; 192 kbps is a reasonable default for music and 128 kbps is enough for speech. Variable quality V0 to V6 trades size for quality without a fixed size.
Each video is read fully into memory to be decoded, so a file may be up to 100 MB. The decoded sound must also fit a memory budget, which depends on its length, sample rate and channels: about 11 minutes of 48 kHz stereo on a computer, about 5 on a phone or tablet, and up to 30 minutes of mono at lower rates. Longer sounds are refused with a message. The browser must decode the whole audio track before this can be checked, so a very long video can still run out of memory. Up to 10 videos can be added at once, one at a time. Sound with more than two channels keeps only the front left and right. Chapters, subtitles and tags are not copied. The video is never uploaded: it is read and processed in this tab, and the MP3 encoder (about 0.9 MB, the encoder script and its WebAssembly file) is downloaded only the first time you convert. The encoder is LAME, which is licensed under the LGPL, and the page links to the LAME project.
Add the MP4, choose a bitrate such as 192 kbps, and press Extract audio. The MP3 appears with the video's name. A 3-minute video gives about 4.1 MB at 192 kbps. It all happens in this tab, so no upload is needed and the video stays on your device.
Either the video has no audio track, or its audio uses a codec your browser cannot play, which is more likely with MKV, AVI or AC-3 audio. Try an MP4 or MOV with AAC audio, or open the page in Chrome or Edge. A file with copy protection cannot be read.
For a spoken lecture, 128 kbps is clear and small: a 30-minute lecture is about 28,800,000 bytes, about 27 MB (an estimate of the audio data; the file is slightly larger). For music, 192 kbps or 256 kbps keeps more. The tool allows 128, 192, 256 and 320 kbps, or variable quality.
Yes. A video may be up to 100 MB. Its decoded sound must fit a memory budget: about 11 minutes of 48 kHz stereo on a computer, about 5 on a phone or tablet, and up to 30 minutes of mono at lower rates. Up to 10 videos can be added at once. The limit exists because the whole file and its decoded sound are held in memory. Cut a longer video into parts first.
It keeps what the audio track had, compressed again. Most video audio is already AAC, so converting adds a second round of compression. At 192 kbps or higher the extra loss is small for most listening, and the original video is not changed.
No. Your browser decodes the sound and the encoder runs in a background worker in this tab, so no server receives the video. The page counts visits with Google Analytics using rough groups such as 10 to 100 MB, never a file name or a video. Your bitrate and channel choices are remembered on this device.