Cut a long voice recording into numbered chunks at its pauses with ffmpeg on Windows and check…

Guide ID: art_01M7CCQSVEF62HPA3RNYWDQNDA

Editorial publication checks are separate from community and evidence review. The evidence labels retain this revision’s reported inspection, execution and review status.

Guide title

Cut a long voice recording into numbered chunks at its pauses with ffmpeg on Windows and check that nothing was lost

Coverage

For people who have one long recording (an interview, a lecture, a voice-over take) and a tool that only takes short clips. A small script finds the quiet stretches, plans the cuts in the middle of them, fills each piece as far as your limit allows and writes numbered WAV files, then re-reads the pieces to check that nothing went missing. A dry run shows the plan first and writes nothing. Honest limit: tested only on invented tones and noise, not on real voices, so the pause settings are a starting point and the cuts need a listen.

Example

On a made-up recording of about 24 seconds with four pauses, the plan gave four pieces and every cut fell inside a pause; joined together, the pieces were identical to the source. With no pause at all, the script cut at the limit and said that it might have gone through a word. A ten-minute test file was split into 23 pieces in about 14 seconds.

Useful starting points

• Run the dry run first and read the notices before you cut.
• If the background is noisy, raise the pause level a little and plan again.
• Listen to the pieces on both sides of any cut that was not in a pause.
• Listen to a chunk that is reported as silent before you delete it: quiet speech can look silent if the level is set too high.