FFmpeg audio recipes Run each command from a folder containing the named input file. Each output gets a different filename. If FFmpeg prompts to overwrite an existing output, keep it or choose a new destination. 1. Extract the first audio stream from the MP4 and decode to a 48 kHz PCM WAV. ffmpeg -i two-clap-reference.mp4 -map 0:a:0 -vn -ar 48000 -c:a pcm_s16le soundtrack.wav 2. Copy the first audio stream from this AAC-in-MP4 fixture to M4A. ffmpeg -i two-clap-reference.mp4 -map 0:a:0 -vn -c:a copy original-audio.m4a The input codec must fit the destination container. Consult ffprobe first. 3. Split a stereo WAV into actual left and right mono channels. ffmpeg -i interview-stereo.wav -filter_complex '[0:a]channelsplit=channel_layout=stereo[left][right]' -map '[left]' -c:a pcm_s16le left.wav -map '[right]' -c:a pcm_s16le right.wav This does not separate sources mixed onto both channels. 4. Decode a 2-second excerpt from the 1-second mark and apply a 100 Hz high-pass. ffmpeg -ss 1 -i steady-hum-original.wav -t 2 -af 'highpass=f=100' -c:a pcm_s16le high-pass.wav This attenuates low frequencies in everything. Listen for changes to speech. Examine any output with ffprobe. ffprobe -v error -show_entries stream=codec_name,sample_rate,channels,duration -of json soundtrack.wav Fixture files and observed results: https://www.saturalabs.com/audio-workflow-kits/ffmpeg-recipes