In this issue's "As We See It," Wes Phillips describes his reaction to comparing the same music recorded at the 96kHz and 44.1kHz sample rates using the dCS Elgar D/A for playback. (For playback, two AES/EBU cables connect the Nagra's digital outputs to two of the digital inputs of a dCS Elgar processor; when this detects the two 48kHz two-channel datastreams, it remultiplexes each into a 96kHz single-channel stream.) Having been able to carry out similar comparisons with the two sets of Rhapsody cardioid master tapes, I am astonished at the difference. The response of the B&K microphones we used for Rhapsody is about 10dB down at 30kHz, while tweeters rarely extend much above 30kHz. And my hearing cuts off above 16kHz! You would think, therefore, that extending the recorded bandwidth by just over an octave, from 22kHz to 48kHz, would be irrelevant.

Yet in level-matched comparisons the 96kHz-sampled tapes had a palpability, an ease to their sound, that the 44.1k versions, as good as I thought they sounded in isolation, only hinted at. The clarinet in Rhapsody in Blue, for example, sounded more like an instrument with a pipe of air vibrated by a reed and the player's embouchure. And the sense of the acoustic space captured on tape was more tangible. There is no doubt in my mind that the higher sampling rate gives a sound more true to the mike feed. A telling indicator was the deflated look on Hyperion Knight's face when I switched back to a 44.1k tape after listening for a while to a 96k tape while we were deciding what takes to use for the Rhapsody master. If you only hear the CD-standard 44.1kHz sound, you won't be aware of what you're missing. But listen for a while to the same music with the higher sampling rate and you really resent it being taken away.

It is naÏve, I believe, to attribute this improvement solely to the increase in bandwidth. Yes, there is published research showing that humans react physiologically to frequencies above the nominal 20-20k band regarded as "human hearing." And it has been postulated that while we don't "hear" ultrasonic frequencies, they do contribute to the envelope of a waveform in a manner that we can detect. While the dCS converters for both sampling rates are almost identical, the high-speed dCS 902 must be somewhat different in that any artifacts from its anti-aliasing and decimation filters will be either an octave higher in frequency or will last less than half the time, however you wish to look at the matter. This surely must be a contributor to the difference in sound quality.

But yes, whatever the reason, I believe the difference due to doubling the recording bandwidth to be significant. If you're attending HI-FI '97, taking place at San Francisco's Westin St. Francis Hotel from May 29 to June 1, you'll be able to hear the difference for yourself in the Canorus room.

I can't wait for the high-quality audio DVD to appear.—John Atkinson