Polyphony

[sbrbot]

On Yamaha site for some CVP models it is stated that polyphony is 256 and somewhere 128+128. What's the difference?

[dbjorck]

Hi!

It's a quite clever thing actually.

256 means you have all 256 on one sound chip/card.

128+128 means you have two sound chips with 128 each. The OS then allocates one of them to the voice most likely to run out of polyphony, usually piano due to the pedaling, and the other voices, percussion or whatever, to the other.

Thanks to the intelligence of this allocation it makes for a much smoother experience, like you actually have even more than 256. With 256 you can run into the piano sound starting to clip if you use the pedal and have a lot of other voices, not so much with 128+128.

Brgds

Danny

[sbrbot]

With 256 you can run into the piano sound starting to clip if you use the pedal and have a lot of other voices, not so much with 128+128.
Sorry but it seems that I didn't get this, why would sound start to clip in case of 256 polyphnoy, one processor is overloaded with MIDI data for processing, or ... ?

[dbjorck]

Hi!

You can think of it as a PC with dual core CPU instead of one. But that's not the main point here. Your problem may be with the usual misconception that simultaneous notes=polyphony. That would indeed be the MIDI view of it, but that's not what's happening here. It's the same amount of notes/events, but could clip in the first solution but not the second. What's so easy to forget about polyphony is that once the note is finished, it continues to sound. We take that for granted on acoustic instruments because it is simply an echo within the body of the instrument. But that means the sound chip in an electronic instrument has to continue sounding - after the note finished. From a MIDI point of view there are no more events, or possibly just some sustain pedal events. But the processing needed to continue the sound is quite a lot. That's why a second chip is actually very clever - for the post sound effects. It's kind of difficult to explain, but does that help?

Brgds

Danny

[sbrbot]

Here is one Yamaha CVP-407/409 review (http://www.yamahamusician.com/clavinova_cvp-407-409.html) saying something about this concept 128+128 polyphony.

[dbjorck]

Hi!

I've heard many versions of this. Here they are theorizing that one is dedicated to AWM and the other to styles; that doesn't make sense to me since the styles use AWM sampled voices as well. Another version of this theory that I've heard before, is that one would be dedicated for XG voices and the other for non XG. That makes more sense. But what I read (forgotten where), is what to me makes most sense and what I tried to describe above;

The OS decides in real time which chip to use, depending on what they are already doing; ie basically it selects the one least in use. And also according to what I read, it prioritises voices with more effects and tries to single out the most demanding voice in use and dedicates it to one chip. But if another demanding voice comes along, it will still not "kill" it, but rearrange again according to which chip is most available. So, simply put, it is like a two processor PC with paralel processes and the OS decides which processor does what based on priorities and activity levels. This may at some times turn out to be GP1 running on one chip, and a lot of XG voices on the other, but not necessarily. Add GP2 and they may run on the same chip. Add another XG voice or whatever, and it may run together with the GP depending on how busy the other chip is with the less demanding XG voices.

I do think it is much smarter than hard coded limitations like suggested. It would mean that if you for instance only played XG voices, you were limited to 128. I do not think that's the case.

Brgds

Danny