Split a stereo or 5.1 file into one mono WAV per channel with ffmpeg on Windows: why the plain…

Guide ID: art_01M7BMY5GHFA42DHX1QJ2K2NTV

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

Split a stereo or 5.1 file into one mono WAV per channel with ffmpeg on Windows: why the plain channelsplit command writes 16-bit files and leaves empty ones behind, and a script that checks every channel file sample for sample

Coverage

For people who have a stereo or multichannel file and need one mono file per channel without losing bit depth. A PowerShell script splits the first audio stream with channelsplit, writes one WAV per channel in the source's sample format (16, 24 or 32-bit integer, or float), compares each file with its source channel by MD5 of the samples, and notes silent or identical channels. It ran on generated tone files (stereo, 5.1, FLAC, AAC in MP4) with ffmpeg 9.0.2, including eight refusal cases; real recordings and listening were not tried.

Example

A plain channelsplit command on a 24-bit stereo WAV wrote 16-bit files whose samples differed from the source; naming pcm_s24le on each output gave identical samples. An unmapped output stopped ffmpeg with 'unconnected' and left empty files, and a wrong layout name (5.1 for a 5.1(side) file) was accepted silently.

Useful starting points

• Name the codec of every output (pcm_s24le for 24-bit); the WAV default wrote 16-bit files here.
• Map every output of channelsplit, and delete the empty files left by a failed command.
• Compare each channel file with its source channel by MD5 of the samples decoded at full depth; the md5 muxer writes 16-bit sound unless you name a codec.
• A channel that is silent or identical to another is worth knowing before you edit.