Video files contain two separate streams: a video stream (the images) and an audio stream (the sound). These streams are multiplexed inside a container format (MP4, MKV, WEBM). Extracting audio means pulling the audio stream out of the container and saving it as a standalone audio file.
This is not the same as "converting" the video to audio. A proper extraction copies the audio stream without re-encoding it, preserving the original quality bit-for-bit. The result is smaller than the video file because you discard all the visual data.
Common reasons to extract audio
Lecture recordings: A professor recorded a 90-minute lecture as a video. You want to listen during your commute. The video is 1.2 GB. The extracted audio is 85 MB. It fits on your phone and plays in any podcast app.
Podcast episodes published as video: Many podcasts publish video versions on YouTube. If you prefer audio-only, extracting saves bandwidth and storage.
Music from personal recordings: You recorded a live performance on your phone. The video is shaky and dark, but the audio captured the music well. Extract the audio and discard the video.
Transcription workflows: Speech-to-text services accept audio files. Uploading a 2 GB video file when only the audio matters wastes time and bandwidth. Extract the audio first, then transcribe.
Output format options
The extracted audio format depends on what was in the original video:
MP4 videos typically contain AAC audio. Extracting without re-encoding gives you an M4A or AAC file. WEBM videos typically contain Opus or Vorbis audio. Extracting gives you an OGG or OPUS file. MKV videos can contain almost anything: AAC, MP3, FLAC, AC3, DTS.
If you want a specific output format (like MP3), the audio must be re-encoded from the source format to MP3. This introduces one generation of compression but gives you universal compatibility since MP3 plays on every device ever made.
Quality considerations
Extracting without re-encoding: Zero quality loss. The audio bytes are copied exactly as they exist in the video file. This is the fastest option and should be your default choice.
Extracting with re-encoding to MP3: Slight quality loss from transcoding. If the source audio is 128 kbps AAC and you encode to 192 kbps MP3, the quality difference is inaudible in practice. If the source is 320 kbps AAC and you encode to 128 kbps MP3, the quality reduction is noticeable on good headphones.
Rule of thumb: encode to a bitrate equal to or higher than the source for transparent quality.
File size expectations
Audio is dramatically smaller than video. Here are typical sizes for extracted audio:
- 1-minute video at standard quality: Video ~15 MB, Audio ~1 MB (AAC 128 kbps)
- 10-minute video: Video ~150 MB, Audio ~10 MB
- 1-hour lecture: Video ~900 MB, Audio ~55 MB (AAC 128 kbps)
- 1-hour lecture as MP3: ~55 MB at 128 kbps, ~110 MB at 256 kbps
The audio is typically 5-10% of the total video file size. The MediaBrew extract audio tool reads the audio stream from the video container and saves it as a standalone file in your preferred format (MP3, AAC, WAV, or the original codec). Processing runs locally via FFmpeg WebAssembly.