Reprinted with permission from Pro Audio Review Magazine

If the Compact Disc had a 48kHz sampling rate 96kHz would make more sense, but it doesn’t, so why 96k? To my knowledge no one has come up with a graceful way to convert from 96kHz to 44.1kHz without a lot of number crunching and a big sonic penalty. Converting from 88.2kHz to 44.1kHz on the other hand is a much easier task and guess what, it sounds better. There is more unfounded hype about 24/96 than just about anything I can recall since the beginning of the digital ice age (love those puns). Some would like to see the CD as we know it today go away, thereby not having to worry about 44.1. But lets face it, like it or not we are stuck with 44.1 and it is not bound to go away any time soon. In fact 99.9 per cent of the people who listen to CD’s think they sound just fine, and more than likely couldn’t hear the difference between 44.1kHz and 96kHz. So that raises the next question, is 96kHz enough of an improvement? If we go under the assumption that 99.9 per cent of the people can’t hear the difference, I guess the answer is no. The next question would be how high must we raise the sampling frequency so that a larger percentage of people could hear the difference or better yet enjoy the sonic benefits? Is it 192kHz or 384kHz? How far do we go?

Heading Down the Wrong Path

I have been a big proponent of higher quality digital audio having used converters that output more than 16-bits and recorders that store more than a 16-bit word for some years now. I think anyone would agree, as the word length increases the quality improves. However the audible increase in quality from 16 to 20 bits is greater than the increase from 20 to 24 bits. And as you raise the sample rate the quality increases as well, also at a disproportional rate. That is, going from 44.1kHz to say 50kHz can be a bigger quality improvement than going from 50kHz to 96kHz.

Having worked with many high density PCM formats I consistently come away from projects or listening tests not totally satisfied. When compared to an analog mic feed or the buss of a really good analog mixer there is something that just isn’t right regardless of the sample rate or word length. Having said that, I am coming to the conclusion that PCM is fundamentally flawed. Defining what is wrong is not an easy task but I have to think that is has at least in part to do with the alignment of fundamental frequencies with their associated harmonics. It is fairly well known that antialiasing or brick wall filters have lots of phase shift. This results in timing or phase errors in the audio pass band and consequently have an effect on high frequency harmonics of both natural sounds and musical instruments. Raising the sampling rate and moving the corner frequencies of these filters upward is definitely a step in the right direction but does not eliminate the problem. The very fact that many artist producers and engineers still prefer to record to analog tape with all of it’s problems and compromises reinforces my belief that something is inherently wrong with PCM.

One Bit to the Rescue

Several years ago Sony engineers began exploring ways of archiving the vast quantity of deteriorating masters in vaults of the record labels. Many of these masters are historical treasures and are literally falling apart. Consequently the goals were to transfer these aging masters to a format that was more stable and be able to easily convert at a later time to almost any new release format not yet defined. Of primary importance was to come up with a technique that could preserve without adding or taking away from the original sound. This made it necessary to think somewhat from scratch since PCM did not satisfy some of the goals.

One bit recording technology has been around for a while but has never made it to market as a product for some reason. Direct Stream Digital (DSD) is a new name for one bit or bit stream coined by both Philips and Sony. In fact in some ways it is a bit (can’t help these puns) like going part way back to analog. There are no decimation filters or defining of word length or any interpolation filters for that matter. The process is a bit stream representation of the analog wave form thereby making a much closer resemblance of the input.

The current version of DSD has a sample rate of 2.8224MHz (which is 64 times 44.1kHz). this frequency was logically chosen so that a simple down conversion to either 88.2kHz or 44.1kHz could be made. Sony has also developed a new super down converter called SBM Direct, not to be confused with Sony’s Super Bit Mapping. I have personally used this device in the mastering of my DSD original recordings and find it amazing that a good part of the wonderful sonic picture of pure DSD can be realized in a 16-bit/44.1 package.

There are approximately 650 million CD players in homes around the world. It makes a lot more sense to me to continue working to make the current CD sound better with new pro audio tools while not rushing into a new consumer format that still has inherent limitations.

Next it’s off to Sweden to record the Stockholm Jazz Orchestra in DSD surround sound. This project marks the seventh album for me using DSD. Any one want to buy some slightly used PCM gear?