Synclavier Regen User Manual
27
outputs. This is a simple test, but it proves invaluable to track down audio
problems if they do occur in your signal chain.
Humans can only hear up to 20KHz sounds in the prime of life and it
deteriorates as we age. Nyquist theory says we only need a sample rate double
20KHz to reproduce the spectrum of human hearing, so CD’s 44.1KHz or
Regen’s 50KHz is more than adequate. However, there are notable reasons why
we may want to use a higher than Nyquist sample rate. Nyquist’s
mathematical theorem assumes a perfect reconstruction filter and there is no
such thing as a perfect filter. The higher the sample rate, the more the
specification of the reconstruction filter can be relaxed. This is especially true
when it comes to sigma-delta DACs found in most modern audio convertors.
Most high-end DAC architectures are aware of this, and select a combination of
digital and analog filters to suit the incoming sample rate. Also, many DAC
architectures up-sample the incoming audio, and if that’s going to happen
anyway, if Regen supplies those additional samples they will be truer to the
sound intended than if the DAC interpolates them.
Another reason you may choose to use a higher sample rate is if your
extensively using FM on your sounds. The FM algorithm produces artifacts in
higher frequency registers than an unmodulated wave. Normally, most are
under the Nyquist frequency, but as you increase the amount of FM applied,
you can push beyond it, and those artifacts will “fold over” into the audible
frequency range, and are often undesirable. A higher sample rate will
potentially avoid this effect. NB: You could also adjust this by mapping
Keyboard to FM Amount such that higher keys will reduce the FM Amount
(mapping sources to destinations are covered later in the manual in the
Modulators section).
So why didn’t we use a higher sample rate on Regen? There’s a trade off. The
more audio samples Regen has to create per millisecond, the more it crunches
the CPU. So the higher the sample rate, the fewer voices its cable of producing.
We settled on 50Khz as a good sweet spot where you get lots of voices and
great sound quality.
