Whatever the camera wrote, one audio track out
Most “video to MP3” pages accept MP4 and shrug at everything else, which is unhelpful, because real video arrives in whatever format the device that made it preferred. A phone shoots MOV or MP4. A download is often MKV or WEBM. A camera from ten years ago wrote AVI. A camcorder wrote MTS or M2TS. An old handset wrote 3GP.
This page accepts all of those: .mp4, .m4v, .mov, .mkv, .webm, .avi, .3gp, .mts and .m2ts. Anything outside that list is refused by the dropzone with an unsupported-type message before any processing begins, which is better than a failure five minutes into a run.
What happens next is the same regardless of which one you brought. The video stream is discarded, one audio stream is picked by FFmpeg’s own default selection and decoded, and the result is encoded to MP3. Because video is the overwhelming majority of the bytes in any of these files, the output is dramatically smaller than the input, and the page reports both sizes when it finishes.
The extension picks the demuxer FFmpeg reaches for
There is a small design decision inside this tool that the single-format converters on this site do not need. FFmpeg decides how to parse a file partly from its filename, so a converter that only ever sees one format can hand every file over under one fixed name. This one cannot, because naming an MKV as though it were an MP4 would point the wrong parser at it.
So the module keeps a lookup table of the accepted extensions, lowercases whatever you dropped, and writes the file into FFmpeg’s virtual filesystem under a matching name. The table carries an .mp4 default for anything it does not recognise, but nothing ever reaches that branch in practice, because the dropzone has already refused every file whose extension is not on the accepted list.
Pulling the audio out of a video
- Drop your video onto the box above, or click Choose a file and pick it. The accepted extensions are printed inside the dropzone itself, in brackets on the Drop your file here line.
- Choose an Audio quality. 192 kbps (recommended) is the default and suits nearly everything; take 320 kbps (highest quality) for music.
- Click Convert Video to MP3. A progress bar and a Cancel button replace the button while FFmpeg decodes and encodes.
- Download the MP3, which keeps the video’s filename with the extension swapped.
A video with no audio track, and what that looks like
The output arguments sent to FFmpeg begin with -vn, which drops every video stream so that only audio reaches the encoder. That is exactly what you want, and it has one consequence worth knowing about: if the file has no audio stream either, there is nothing left to write.
Silent screen recordings, muted phone clips and some exported timelapses fall into that category. The run ends with the page’s generic message, Something went wrong while processing your file. Please check the file is valid and try again., because no output file was ever produced for it to hand back. It is not a bug and no bitrate setting changes it. If a video genuinely should have sound and does not, the problem happened at recording time.
Always a re-encode, never a stream copy
Some videos already carry MP3 audio, particularly older AVI files, and in principle the audio could be copied out untouched. This tool does not do that. It always decodes and re-encodes through libmp3lame at the bitrate you chose.
The honest cost is a second lossy pass. Video audio is almost always AAC or Opus already, so a re-encode is unavoidable in the normal case anyway, and a consistent, predictable output is worth more than saving a generation of quality in the minority of files where a copy would have worked. It does mean that extracting audio at 320 kbps from a video whose soundtrack was 96 kbps AAC produces a large file carrying a small amount of information.
If you want to keep the video as well as the audio, converting the container is a separate job: MKV to MP4 tries a stream copy first and falls back to an H.264 and AAC re-encode when the copy fails or when you pick that mode yourself, and the video converter and compression tools guide covers the rest of that side.
MTS, M2TS and the camcorder cases
The two extensions people least expect to work are the AVCHD ones. Consumer camcorders from roughly 2007 onwards wrote MPEG transport streams to their internal cards, split across numbered files in a BDMV or AVCHD folder structure, usually with Dolby Digital audio alongside H.264 video.
Those files are supported here directly, which matters because they are frequently the last surviving copy of a wedding, a recital or a family holiday, and the software bundled with the camera stopped running several operating systems ago. The audio comes out as a normal MP3 that anything will play.
Transport streams are also large, and the whole file has to be read into the tab before decoding starts. If a long recording fails partway, memory is almost always the reason, and nothing here trims an MTS directly: Trim Video takes MP4, MOV, M4V and WEBM only. MTS to MP4 rewraps the stream into a container that trimmer accepts, usually with a stream copy rather than a re-encode.
Bitrate for a lecture against a concert recording
The three rungs map to one FFmpeg argument each. 128 kbps (smaller file) is right for speech: lectures, meetings, webinars, interviews, anything where the content is a person talking and the file is going to a phone or a transcription service. 192 kbps (recommended) is the default and handles mixed material without thinking about it.
320 kbps (highest quality) is worth it when the soundtrack is music you intend to listen to properly, such as a concert recording or a music video, and it is wasted when the source audio was heavily compressed to begin with. If your file happens to be an MP4 and you landed here by accident, MP4 to MP3 does exactly the same thing from a page written for that one case, and the wider audio tools handle trimming and levelling once you have the MP3.