Recording streaming audio 4934

On Tue, 04 Oct 2005 18:39:45 +0000, spike1

This is the approach I have used for some time, though this command format is a slight change from older versions of mplayer.

Generally the downloaded .ram file is a container which holds the actual stream url, I prefer to listen in xmms, so throw all the steps into a shell script which will extract the address and invoke lame taking input from a pipe, the full script looks like this

#!-bin-bash

for i in `ls -1 *.ram` do streamurl=`cat $i`

if -e $i.mp3 ; then echo "-Warning- output file $i.mp3 already exists, renaming old copy" mv $i.mp3 $i-`date +%s`.mp3 fi

lame -t -b 96 -m j --quiettmp-myfifo.wav $i.mp3 & mplayer -cache 1024 -ao pcm:file=-tmp-myfifo.wav $streamurl RC=$?

if "$RC" = "0" ; then if -e $i.mp3 ; then mv $i $i.done fi else

echo "-Warning- mplayer exited bad" fi done

Before using the script create the desired pipe with mkfifotmp-myfifo.wav

This will process a batch of .ram files in the current directory and generate corresponding mp3's, it's a good idea to give meaningful names and dates to the .ram files you download, so the created mp3's are easily identifiable. Different source material may not justify a bit-rate of 96kbps or joint-stereo.

Some issues remain however:

If mplayer exits bad (e.g. user hit's ctrl-c), the instance of lame invoked in that iteration remains running, it needs to kill that specific instance of lame in that event, and not any-all running copies of lame owned by that user

It runs in real time, which isn't so bad if you want to begin listening immediately, but is time-consuming if capturing a series or number of files or other activities affect available bandwidth with the possible risk of network errors, though the cache helps somewhat.