Hi,

Currently I record mic input like this
However, I was wondering if I could split the mic stereo input in two mono files on the fly via the terminal (no X installed)? Doesn't need to be ogg. mp3 is fine as well.

Has someone an idea?

Thank you,
Ralf

okay, my idea is a little on the crash side, but it just might work.

i haven't implemented a full implementation but i have implemented a basic framework that you an adjust to your needs.

basically the idea is as follows. arrows indicate pipes and the numbers indicate processes

1) arecord (from mix in a format ffmpeg understands) ----> tee left ----> tee right

2) cat left ----> ffmpeg (throw away right and maybe compress)

3) cat right ----> ffmpeg (throw away left and maybe compress)

left and right are both fifos created using mkfifo.

basically i tested this in the following manner:
- created left and right fifos using mknod
- created single file called containing "ababababababababbbababbabab" to represent the left and right audio data
- created script that starts both left and right fifo readers that will "encode" the data.
- streamed "audio data" to both left and right fifos using cat.

--- file "stream" ---
aaabababababbbabababbabababababbababbababababbababa
--- end file ---

--- file "readlr.sh" ---
#!/bin/bash

cat left | sed -e "s/[^a]//g" &
cat right | sed -e "s/[^b]//g"
--- end file ---

in one console i ran readlr.sh first
$readlr.sh

and then in second console i ran the "streamer"
$ cat stream | tee left | tee right

the output on the reader console was as follows
$ ./readlr
aaaaaaaaaaaaaaaaaaaaaaaa
bbbbbbbbbbbbbbbbbbbbbbbbbbb

how you will split the left from the right is up to you. i would probably use ffmpeg and try to eliminate a channel using channel mappings and setting the max number of channels to one. i don't know if this will work as expected but that's where i would begin. you might find that ffmpeg will just downmix both channels into one which is obvious not what you want.

if you get this working, please post a solution as i'm sure others would appreciate that.

good luck.

$ sox stereo_input.wav output_left.wav mixer -l
(avg -l on older versions)

$ sox stereo_input.wav output_right.wav mixer -r
(avg -r on older versions)

Not on the fly. But it's possible. Some applications let you record each channel individually. Just most cli based ones would require you to launch separate applications to record both channels individually at the same time. Leaving you with an unknown latency between channels, if you ever needed to seemlessly rejoin to a stereo image. Sox lets you do the extraction in post, seemlessly, so it really depends on how on the fly you need things to be.