In this guide
Exporting a soundtrack keeps the sounds already mixed into the selected audio track. Use WAV for an editable decoded copy, or copy compatible AAC into M4A to avoid re-encoding it. Isolating a voice or effect within that soundtrack is a separate task.
To extract audio from video, export the video's audio track without its picture. If you need the whole soundtrack, an audio export in an editor or a stream-copy command may be enough. If you need a voice, bell, or instrument separated from other sounds on that track, export alone will not do that.
Choose the output by your next task. A PCM WAV is a useful working copy for an audio edit. A compatible encoded-track copy avoids another lossy encoding pass. Neither creates an original microphone recording from a finished mix.
Choose soundtrack export or sound isolation
| What you need | Starting workflow | What remains mixed together |
|---|---|---|
| All the audio from a video | Export or copy the selected soundtrack | Every sound already mixed on that audio track |
| A short scene's complete sound | Export the soundtrack, then trim the intended interval | All sounds during that interval |
| Only a speaking voice | Try a voice-focused isolation | The estimate can retain background or lose voice detail |
| One effect underneath speech or music | Try sound-effect isolation | The estimate needs a check for bleed and lost attack or tail |
For example, a video can contain one audio track holding dialogue, music, and applause. Exporting that track retains all three. A second selectable audio track, if present, might instead be another language or commentary. A file's channel count and its number of selectable audio streams are also different: one stereo stream already contains two channels.
Export a video soundtrack with Audacity
For a graphical workflow, Audacity's audio import documentation explains that its optional FFmpeg library can import audio from most unprotected video files. Follow its FFmpeg installation guidance for your operating system if the video format needs that support.
- Keep the original video. Work on a separate export rather than replacing your source.
- Choose File → Import → Audio. Select the video. The file-type filter may hide a supported file; review the import dialog's format choices if needed. If several audio streams are offered, choose the intended one and listen to identify it.
- Check the imported audio. Play a complete spoken phrase, a busy section, and the ending. Confirm that you imported the wanted language, commentary, or mix before exporting.
- Export an audio file. Use File → Export Audio and choose the full intended range. For a basic editing copy, choose WAV with signed 16-bit PCM and an appropriate sample rate such as 48 kHz. Check the export's mono/stereo choice instead of assuming it preserves your preferred layout.
The Audacity export reference covers file format, sample rate, channels, and export range. Saving an Audacity project is different from exporting a playable WAV. If your next tool needs mono or stereo, deliberately choose that layout; do not mistake an automatic mixdown for separate microphone tracks.
Copy AAC from MP4 to M4A without re-encoding
If you already use FFmpeg, first inspect the audio streams with its companion FFprobe. These examples run locally after you install those tools; they are not commands sent to Saturalabs. Replace the example filenames with your files, and use a new output filename.
ffprobe -v error -select_streams a -show_entries stream=index,codec_name,channels,sample_rate -of json "input.mp4"When the selected stream is AAC and fits the destination container, the following command copies the first audio stream into M4A:
ffmpeg -n -i "input.mp4" -map 0:a:0 -vn -c:a copy "soundtrack.m4a"-n refuses to overwrite an existing output. -map 0:a:0 selects the first audio stream from the first input. It does not select every soundtrack or choose a particular speaker. -vn excludes video; -c:a copy copies encoded audio packets rather than decoding and re-encoding them. FFmpeg's official command documentation describes stream selection and stream copy.
This AAC-to-M4A example is not a universal container conversion. If the chosen codec cannot be stored in M4A, choose a compatible container or intentionally convert it. Changing a filename from .mp4 to .mp3 does neither. Container metadata and timestamps can change during a copy even when the audio packet data is preserved.
Convert the selected soundtrack to WAV
For a decoded 48 kHz, 16-bit PCM WAV, use:
ffmpeg -n -i "input.mp4" -map 0:a:0 -vn -ar 48000 -c:a pcm_s16le "soundtrack.wav"This converts the first audio stream rather than copying its original encoded bytes. It does not deliberately request a mono or stereo downmix. Inspect the resulting channel layout: multichannel sources can produce multichannel WAVs, which Saturalabs' local preparation tools do not accept.
We checked the extraction recipes on original synthetic video fixtures. An AAC-to-M4A copy retained the selected stream's encoded packet hashes. A WAV conversion produced audio-only 48 kHz, 16-bit PCM, and its decoded PCM matched a direct decode of that selected source stream. Those checks validate the tested recipes, not every codec, damaged container, player, or editor interface.
A WAV can be much larger than the video file's compressed audio. For mono or stereo PCM at the same rate, 16-bit samples use two bytes per channel per frame. The audio file size calculator estimates the resulting payload and a basic header. A larger WAV does not recover detail previously discarded by AAC or another lossy codec.
Check timing before replacing the video's audio
Inspect the extracted audio at the beginning and end. Containers can store stream start offsets or codec timing information; do not assume a newly exported file retains every original timestamp relationship. If the audio starts later than the picture, preserve that placement deliberately in the editor.
For an excerpt, keep its original start time alongside the filename. The free audio trimmer can shorten a supported short MP3/WAV working copy, but it cannot identify where that excerpt belongs in another video. Follow the audio replacement and sync guide when returning it to the timeline.
Mute the corresponding original sound when auditioning a replacement, then check a clear word or action near both ends. Listen to the final video export too. Two overlapping copies can sound louder or phasey even when each audio file is fine on its own.
Common questions
Can I extract sound from an MP4 with no audio stream?
No soundtrack can be exported if the file has no audio stream. Inspect it before interpreting an extraction error as a volume problem. A present but silent audio track is different from a file with no audio track at all.
Will extracting audio remove background music?
No. Whole-track export retains music already mixed with the dialogue. For that task, follow the music removal and voice-preservation guide. Source isolation estimates parts of a mix and requires a listening check.
Do I need to export audio before using Saturalabs?
No. The AI isolation upload accepts MP4 and MOV directly, as well as MP3 and WAV, up to 50 MiB (52.43 MB). It processes one described source and returns separated audio. AI downloads require an account and sufficient credits. Use the extraction workflow here when you need the complete soundtrack or a working file for another editor.
Can a browser always decode my MP4 or MOV?
No. The extension identifies a container, not a guarantee that its audio codec and layout work in a particular decoder. If a local browser tool rejects it, use an editor or converter that supports that file rather than changing the extension. The MP3 and WAV editing guide helps you choose the next working format.