[CLUE-Tech] cdparanoia and ripping CDs with Linux
Grant Johnson
grant at amadensor.com
Wed Mar 20 08:55:13 MST 2002
CD's are sampled at 44.1Khz. This means that everything that happens in
between these 44,100 samples per second is lost. Yes, there is a lot
happening. On a much less serious note, it is 16 bit samples. This
means that there are only 2^16 places for that value to be. Any more
detail than that is lost. The biggest advantage of 20 bit audio over 16
is that you are able to use the extra room to increase dynamic range,
and increase the level of the signal over the level of the noise.
I am an audio engineer (was by trade, now by hobby) and have a recording
studio in my basement (well, it is in Scott's basement now, but it will
be in mine at the end of the month).
Digital has loss. Analog has noise. Pick.
As for MP3, it is a high logg compression, which means that they give up
quality for size, but it is also 1/10 the size of wav. Wav can be
compressed with either a lossless compression, or even a lossy
compression, however, most of the time it is uncompressed data, which is
what cdparanoia makes.
I have a new audio compression algorithm I have designed, but have not
had the time to write the C to do it. If anyone out there is a C jockey
with an interest in audio and a touch for bit twiddling, tell me, and we
can work on this. Based on my preliminary calculations, we should get
about a 1/6 size file (compared to uncompressed 44.1Khz 16 bit) with no
loss at all. This will be by doing an initial pass which should get a
4:1 ratio, followed by traditional file compression.
Traditional file compression like gzip does not work well with audio
because they work on compressing paterns, and audio does not appear to
have those paterns from a data standpoint.
More information about the clue-tech
mailing list