[Rich[_2_]]

Dave wrote:
I have been disappointed with the audio quality of some CDs I have
bought recently. Is there a free program I can use to get an accepted
measurement of fidelity? (like a signal to noise ratio)

Some famous commercial recordings have a low s/n being recorded at
home. I had assumed my cassette tape (forget which song now) was
bad till years later when I bought a CD and it sounded the same.

A few issues ago Maximum PC did a comparison among lossless codecs
and at 192 kb/s there was only one track where any difference could
be heard. Now mp3 is a lossy codec, but at higher bitrates I doubt
there is much difference in the sound. For mp3's I suspect that
256 kb/s is where you'll have difficulty telling the difference.

I have my suspicious that some may have been stored an MP3s and then
"unripped" in the factory.

Long ago I made a custom CD for my wife's aunt and I ripped wav files
from her CD's and remastered the tracks she wanted. She was very pleased
with the results.

Less long ago she wanted another CD. I had forgotten where I put them
and when I found them this time I ripped mp3's at 128 kb/s and remastered
the CD, she was very disappointed with the quality this time. As I recall
I needed to free up some disk space at that time so this seemed quickest.
But obviously mp3's are not as good as the original wav files.

So how can I tell for certain if my CD has
been an MP3, or other lossy format? I'd hope mp3 storage would leave
different markers than the original tape, for example.

There may be another way, there is a public database of CD's, immdb
or something like that. I'm not sure how they identify the CD, perhaps
the volume name, but I suggest that any casually remastered CD, or
any made from mp3's from the net won't have the same volume name, or
whatever they use. That is, I suggest you rip the tracks from the
CD with some software that can identify the disk from immdb, if it
cannot identify the CD, there's a good chance it's as you suspect.
Unless you've for some obscure latin CD or something, the database
is not complete, but for any popular release this should work. I
think Nero will do this, I've not done it for years though.

Note, you don't have to rip the tracks, just see if the CD can
be identified.

Cheers,

Rich


To get a good measure I'd expect some Fourier transforms and signal
analysis to be done, so this should be relevant to sci.physics.