How to record input audio stream with sox rec, pipe it to ffmpeg and split resulting outfiles
AUDIODRIVER=alsa rec -q -b 16 -r 44100 -t wav - silence 1 0.0 0% 1 20.0 3%| ffmpeg -f wav -i pipe: -f segment -segment_time 3600 -c copy /share/test%02d.wav
Translates to:
AUDIODRIVER=alsa --> tells rec which driver to use in case more drivers are available on the system
rec .... --> tell it to shut up (-q; we check recording levels beforehand carefully and now only want to see the screen output of ffmpeg), specify bitrate, resolution and type of the "file" (we pipe out a stream!) to be recorded
NOTE the " - " after "-t wav" THIS tells rec to pipe the stream outward
silence ... --> START recording immediately (0.0 seconds after level is over 0%) and STOP recording once level falls 20.0 seconds under 3%
At the receiving end of the pipe we tell ffmpeg to write out a wav file, take input from pipe and write in segments of 1 hour to a 2-digit-test.wav file (on a smb share or easy access later on)
This should come in handy for long recording sessions that ends automagically once level drops and have "not too long" output files for post-processing.
post-pro could look like:
We start a new ssh session and on the remote we "screen -S my_recording_session", fire the rec command, DETACH the session via CTRL+ a d and close the ssh session. After recording is finished we grab the files off the share and do post processing (cut silence at beginning, normalise to -2 db, mark noise in audacity, export multiple to .mp3).