Replace the sound of a video with a new voice or music file with ffmpeg on Windows: keep the…

Guide ID: art_01M797NNR3TNJT9ASGA2AGS9CB

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

Replace the sound of a video with a new voice or music file with ffmpeg on Windows: keep the picture byte-identical, fit the new sound to the picture's length, and check where the sound starts and that its level did not change

Coverage

For anyone with a finished video and a separate voice-over or music file who wants the new sound in the video in place of the old one. A PowerShell script copies the picture untouched, makes the new sound exactly as long as the picture (cut, padded with silence or delayed), encodes it as AAC, proves the picture packets are identical, and measures where the sound starts and its level in the result. It ran on generated files with ffmpeg 9.0.2, including sixteen refusal cases, and refuses videos it cannot handle safely instead of dropping subtitles.

Example

With a 14 s test clip and a spoken-sentence WAV, the result held the same 350 picture packets and a 14.000 s mono AAC sound starting 3.121 s in, at the same -24.3 dB level. Plain ffmpeg exited with code 0 each time yet kept the old sound, cut the picture to 7.96 s, made a 20 s file for a 14 s picture or put raw PCM in the MP4.

Useful starting points

• Name the streams with -map; without it ffmpeg can silently keep the old sound.
• Fit the new sound to the picture length with apad and atrim instead of -shortest, which cuts the picture when the sound is short.
• Check the picture packets, where the sound starts and its level in the result, then listen to the start, the middle and the end.