The sound in a video is a separate track
A video file is a container (MP4, MOV, MKV, WebM, TS) that holds a picture stream and one or more sound streams side by side. Phones and most cameras store the sound as AAC at 96–256 kbps; screen recorders in the browser and OBS use Opus; cinema rips in MKV often carry Dolby Digital (AC-3) or FLAC; professional cameras record uncompressed PCM.
Because the tracks are separate, the sound can be taken out without touching the picture. This tool reads only the sound packets: a 2.08 GB MOV with 7.5 minutes of 160 kbps AAC gave a 9.09 MB audio file in 1.7 seconds in Chrome, and the video frames were never decoded, so the video codec (H.264, HEVC, AV1) does not matter.
Original audio or MP3: which should you download?
"Original audio" copies the sound packets as they are: AAC becomes .m4a, Opus .ogg, Vorbis .webm, FLAC .flac, PCM .wav and MP3 stays .mp3. Nothing is re-encoded, so the quality is exactly the video's and the job takes seconds even for an hour-long file. M4A plays on iPhone, Android, Windows and Mac.
Choose MP3 when a car stereo, older player or app accepts only MP3. AAC to MP3 is a second lossy compression: it can never sound better than the source, so a bitrate at or slightly above the source (usually 128–192 kbps) is enough. WAV and FLAC are for editing; they keep the decoded sound without further loss but are 5–10 times larger than the original. In MP3, WAV and FLAC mode the original audio is still offered as a second download.
Supported formats and codecs
Containers: MP4, M4V, MOV (including iPhone HEVC videos), MKV, WebM, MPEG-TS/M2TS and Ogg. Sound codecs: AAC, Opus, Vorbis, MP3, FLAC and PCM convert to anything. Dolby Digital (AC-3 and E-AC-3) can be extracted as the original .mka, but no browser can decode it, so it cannot become MP3 here.
Not supported: AVI, FLV, WMV/ASF and old 3GP phone videos with AMR sound — browsers have no reader for these containers or codecs, and DTS tracks. The file list says which case applies instead of failing silently. DRM-protected videos from streaming apps cannot be opened.
Videos with several sound tracks, and iPhone spatial audio
Movies often have one track per language, and Zoom or OBS recordings may keep the microphone and the computer sound apart. When a file has more than one sound track, each row shows a track list with language, codec, sample rate and channels; the default track is picked first, and you can switch before extracting.
iPhone 16 and later can record spatial audio (APAC) together with an ordinary stereo AAC track. Browsers cannot read APAC, so the stereo track is used automatically and the result says so. Surround tracks (5.1) are copied with all channels as original audio and mixed down to stereo for MP3.
Extracting only a section
Open "Cut a section" on a row to set a start and end time, for example 1:30 to 2:10; you can play the video's sound and take the current position. Formats accepted: 90, 1:30, 1:30.5 or 1:02:03.
MP3, WAV and FLAC are cut on the exact sample after decoding. Original audio is cut on packet boundaries because packets cannot be split without re-encoding: about 21 ms for 48 kHz AAC, 20 ms for Opus and 24 ms for MP3, so the start may move by that much.
Large files, long videos and privacy
The video is read from your device in small pieces and never uploaded, so there is no daily quota and no file size cap; a 2 GB file was tested. Original audio needs memory only for the sound itself (a 90-minute 128 kbps AAC track is 87 MB).
Converting to MP3, WAV or FLAC decodes the sound. Browsers with WebCodecs audio (Chrome and Edge 94+, Firefox 130+, Safari 26+) decode long videos in chunks, so length is not limited by memory. Older browsers decode the whole track at once: about 23 MB per minute of 48 kHz stereo, so the tool warns above about 43 minutes and stops above about 78 minutes on a computer (about 13 and 30 minutes on a phone) and suggests original audio instead.