I apologize in advance for the naivete of these questions. I'm trying to understand the issues controlling price and weight in the home/pro keyboard markets. I do try to google things out, but sometimes that gets frustrating.

1) Is it true in general that to double polyphony on a synth, you have to double the number of physical oscillators? The price and weight ALWAYS seem to go up precipitously when you jump from 32 to 64 poly.

2) When layering two sounds together, each using 1, 2 or 4 (for Kurzweil) oscillators wouldn't it be possible for them to combine the two tones _mathematically_ so the same number of oscillators are used in total?

I may not know much, but I'm always thinkin!

Rick

Rick,

I will try to answer your questions as best I can:

1. The statement you make would be true for an analog synthesizer, where there are physical oscillators per voice. Today most "analog" keyboards use software simulations of the oscillators, so having more oscillators in an instrument is mainly a matter of putting a faster processor into it. The excuse that most keyboard makers use is that their software designers are not computer engineers, and that is why it takes them a lot of time to adapt new technologies, and because of that their processors are generations behind the leading edge of today's technology.

Since you post this question in the arranger forum, I will add some arranger perspective. Since the arranger users are typically not interested in creating new sounds, most arranger keyboards use ROMPler technology - essentially the sampler, but capable of only playing back the sound data stored in the ROM of the instrument. This results in the sounds which are virtually perfect reproductions of the real instrument sounds, rather than approximations which can be achieved with the synthesizers of old. The polyphony again is determined by the speed of the processor, though the quality of an individual samples is determined by the size of the wave ROM. Still, even with the best quality individual waveforms in ROM, it is often desirable to layer voices for greater realism and better response to key velocity, which is why many instruments use 2 or even 4 layers in their best voices (I think Yamaha is capable of layering up to 8 tones, but I may be wrong).

2. What you are suggesting would be possible, but you would have to store the resulting waveform in memory, and that would constrain the amount of sound data which could be stored. I thought I read somewhere that the Kurzweil has true polyphony of 48 notes. The vast majority of the instruments utilize a much more expedient way of implementing splits by layering of the waveforms in real time (and using up multiple notes of polyphony in the process)

Regards,
Alex

Many thanks, Alex!