Audio Considerations

Menu:

• CD's, what's wrong

CD's, what's wrong ?

It's not my purpose here to fire against digital sound. In fact, I strongly believe in digital not because is better by itself but only because we've never found a truly good way to store analog sounds (or other phenomenons). Digital is easy to store, this is its main and probably unique advantage over analog.
If one day, a good system to store analog is discovered, that time we can forget digital forever ...

Since the first introduction of the Philips/Sony CD, I felt something was wrong, something was missing and is still missing: Emotion. I do not believe in esoterics: I work in computers and have both the feet right on earth !
But if you simply compare a CD to an analog record on a better than average equipment, you can't miss that medium and treble are not as good on CD's and more important "emotion" is lost. To be honest, I should also admit that bass is usually better on CDs ...

When Philips designed this standard, optical storage was only beginning and particularly limited in capacity. They wanted their system to be:
- practical (and it is)
- contains more than 1 hour of music (and it does)
- and finally, but only finally, be of the best quality possible (could be better !).

That's the reason of this standard: 44.1KHz sampling rate with 16bits definition.

What does this mean:

• Sampling rate: the number of measures of the sound level per second (a little bit more than 44'000 per second)
• Definition: 16bits means 65'536 different values possible.

We can hear up to 20'000 Hz, and with this standard, a 20KHz sound will be measured only 2 (!) times per cycle.
It's enough say Philips and Fourrier, I strongly not believe it !!!
If you really want Hi-Fi, the sampling rate should be at least ten times higher, but if they did it in such a way, they could store only one tenth of the duration on a CD (6 minutes, gee, as an old 45rpm, what a progress ...)

Second problem, the resolution: 16bits. It's enough only for strong sounds ! In fact, you have 65'000 different levels only at the highest sound level recorded but, in the reality, all the subtle sounds (small instruments, reverberations, ...) which makes the "emotion" are much, much lower than that and you could be happy to have a hundred or two different levels to measure them !

Try to imagine how well a low 10KHz sound could be reproduced by such a standard ...

For true Hi-Fi, I don't think we should go under 24bits, which give 16mio different levels. We could also imagine a logarithmic scheme: proportionally more levels for small signals and less for high ones. This is nothing else than compression at recording (like Dolby, but this one could be frequency-linear) and expanding during the playing, but this should be only a last option: a linear system is simplest and though a better one.

Actually, the sampling circuits can't certainly achieve such a resolution and at such a high speed but that should be the main goal when you talk about Hi-Fi.

There are several other problems with CD's like the recovery system that "fills" in the blanks or the 44KHz rejection that implies strong (and bad) filtering or oversampling and other jiitter phenomenons.

Conclusion:

The actual format of CD's is nevertheless a big step forward for the "common soup" Hi-Fi, but certainly not for purists.
It has been designed with commercial and technical considerations in mind first, audio quality was only third in their process. They should have just done the opposite: first fix the goal and only after try to find the technology to reach it.
We (the purists and everybody who wants true Hi-Fi) strongly need another format, but please, Mr. Philips or anyone else, don't make a 96KHz/24bits system including both audio and video, go directly to a consistent audio-only format: the next generation of CD ROM's makes it possible to store such a storm of informations !

Back to Top