Over the years I have put together and refined a system
for automatically recording all programs from my two favorite radio
stations- WNYC-FM, the
main National Public Radio
affliate in New York
City, and WQXR, the lone remaining
classical station. It was originally based on a SPARCStation from
Sun Microsystems and is now based on
PC running LinuxTM.
Those who own TiVoTM
or other types of personal video recorder (PVR) systems will have the
least trouble understanding my motivation for putting this sytem
together. This system changes one's listening habits in many of
the ways the TiVo changes one's viewing habits. The main place
where the comparison breaks down is that Lesk Radio captures two
stations, but in my case those two stations supply over 98% of the
I listen to in home and car.
Yes, these programs are available via web streams that can be captured,
but the audio streams sound inferior to the sound I
obtain with my current system. The off-air signal is also much
more reliable than the web streams. Some of the programming is
available in archives on
various web sites but often not for some amount of time - some hours
later for Marketplace, two or three days for locally produced WNYC
programming, and little if any WQXR programming is available. In
contrast I can play a program at any time after
that program broadcast starts. By default programs are retained
for three weeks. A very few are archived permanently.
The current hardware for the system consists of an AMD-based PC running
Ubuntu 10.4 LTS Linux with two Sound Blaster
audio boards along with a rooftop antenna, and two devoted external FM
tuners. The software section is based around an open-source
encoder from the Lame Project,
for pulling raw digitized audio off the audio board, two small C
programs, a script for Korn Shell, and one whopping big crontab with an
entry invoking that shell script for each program to be captured.
In my home environment the recorded programs are played back with an Argosy media player.
If anyone out there is actually intrepid enough to attempt building a
system like this please be aware that the software and programming is
only a fraction of the total effort involved. I encountered
numerous problems in making the system sound decent and reasonably easy
on the ear, problems that did not stem from using MPEG 1 layer 3 mp3
digital audio compression. Most of these problems came in three
areas: getting a clean FM signal out of the ether, delivering clean
audio from the tuners into the line input of the sound cards at the
proper level for optimum results, and artifacts from poor audio
hardware in both the recording and playback PCs.
Coming someday is more detail and a history of how this system has
evolved over the years.
Maintained by Daniel V. Wilson
and last modified 9 October 2011.