So I was trying out a Yamaha DGX505 and a Casio Privia PX-555R at the Guitar Center in Lake Forest, IL (58 kbds on display!) and they both have 32 note polyphony. I wanted to see if this was going to be a huge hassle, so I tested it, and tested it, and ... NOTHING! In no way could I discern any difference between 32 note polyphony and 88 note ... 128 note ... 22,352 note polyphony!!

I layered. I split. I held the damper pedal down for oh, maybe 8 days. If anyone can describe a test I can do that would reveal what's shakin with this issue I'd appreciate it.

Here's one test I did. First, I layered piano and strings. Then I hit the low E flat hard with damper pedal down and did a slow gliss on the white keys from middle C up to the top of the keyboard. I heard every note of the gliss strike, and when I was done, I lifted the damper pedal and the low E flat was still playing. Lifted finger off the E flat it stopped. If I re-depressed the pedal before lifting finger, low E flat continued -- all exactly as per an acoustic piano.

I could not hear any notes drop out from the gliss, though it's possible some CAREFULLY SELECTED notes dropped out. But they must have one heckova selection algorithm.

I also played Chopin's Fantasy Impromptu with piano/strings layered and the damper pedal down always. Sounded terrible, of course, but never heard a hint of a note dropping out -- except one spot that turned out to be a simple issue stemming from my lack of practice on a nonweighted keyboard. (Note was there, I just hadn't hit it hard enough to hear it.)

These results held true for both the Yamaha and the Privia. I think they've solved the polyphony problem.

I just now occurred to me what I'd probably have to do to hear the 32 note limit. I'd have to hit more than 32 notes at a time, like by laying both arms down on the piano. But geez, how often do you need to do that in RL?

You only have ten fingers.

Gents, when the damper pedal is down, there's NO LONGER ANY NEED for the synth to remember the specific keys that have been pressed. When pedal comes up, that's your release. I honestly think the actual SOUND is carried to infinite polyphony with these new machines. (If you think about it, this isn't that hard. Every overtone really has to be available anyway, for just two sophisticated jazz chords held through the pedal with any decent sound sample.)

I'm completely aware this was once a real issue. I remember notes dropping out all the time when I layered strings with piano about five years ago. But apparantly, polyphony ain't what it used to be.

As far as I can tell, there is no issue whatsoever with 32 note polyphony on a performance synth. In fact, as far as I can tell, 20 note poly would suffice.

Comments?

Rick

Rick, polyphony is for real..It can be a big problem..
Each manufacturer has a different way to handle it[priority]..
I have always preferred Roland's design..besides last note priority, certain channels have priority...the drums and bass are one, two on the list.
Some Roland gear , you can select your priority..but in all the models you are assured the bass and drums will not drop out, followed by the main voice[s]..

If you really need to test "Big Time"..Play a SMF[standard midi file] on your keyboard, and layer strings and piano, using your hold pedal play over the top of the SMF.

32 polyphony is terrible in this scenario.
On some brands 126 polyphony[hint] is bad.

The more forgiven brands are Roland and Korg..

My experiences with PSR's[early 32 voice polyphony]..they would mute everything for a few seconds when they drop out..

The Casio 32 voice are also noticeable with dropouts, although they do not mute the board entirely..

The last not priority design on the 62 voice Korgs handle the drop outs better than the 126 voice Yamaha's..and the Roland 64 voice[G1000] handles this even better than the Korg..

Maybe, drop outs do not bother you in your style of play, but if you listen you will hear the sounds dropping out...it is for real..

BTW..the polyphony is a sound generator limitation..the damper/hold pedal down will not allow any "new" voices...
If you hear a sound it is using polyphony..It is not a key pad issue..

Also keep in mind, more detailed sounds use a combination of tones/partials..as an example the Tyros uses 4 tones for many of the better sounds...each of these tones use polyphony[4 tones=4 polyphony for each key depressed]..

Hi Rick,

The Casio piano is only 32 note polyphony when you are using certain voices on the piano. Once you switch to some of the others that are stereo sampled it drops to 16. Imagine now 16 polyphony now layered with strings or something. Not to mention if you use their onboard recorder. I think that the Casio piano is good, but I think that many makers of keyboards are not disclosing the true polyphony of their boards. Technics did the very same thing with at least one of their boards. The P-30 I believe stated it was 32 note Poly and once you switched it to stereo output it dropped to 16. Somewhat dishonest I feel. However, Polyphony is important. You will really notice it on multiple sustained notes not so much on a gliss. The attack and decay are to fast for that to really show the lack of polyphony. Not to mention if you are in a 32 note polyphony mode, 32 notes are quite a few to have played at one time if your just using one voice. Multiple layers or multitracking will eat polyphony very quickly. Since starting out on a miniMOOG and moving through Korg POLY series and Roland Juno and Yamaha DX series and the Clavinova's I can certainly say without hesitation POLYPHONY is a big key to the value of a keyboard.

Best regards,

Dennis L. Almond
aka TwoNuts

[This message has been edited by TwoNuts (edited 04-12-2006).]

Thanks Fran, Dennis. I suspect most all these issues vary from one manufacturer to another. I don't think Yamaha, for example, is going to divulge any trade secrets! Can somebody direct me to a whitepaper or a serious discussion of how polyphony works on modern keyboards?

Yamaha says: "32-note Polyphony -- What's polyphony, you ask? Simply put, polyphony is how many notes a keyboard can reproduce simultaneously." I think they really might mean SIMULTANEOUSLY, as in, notes played in sequence are unaffected -- even with the damper pedal down.

This is 95% going to be a performance synth. I want to be able to throw together simple arrangements, eventually, but I suspect 32 notes (or even 32 stereo sounds) will be sufficient.

From posts on another board on note-dropping-selection algorithms, I now think there's a possibility I really did hear dropped notes in that one spot where I wasn't hitting the note hard enough. So I have a new test idea: I'll hit a middle E flat SOFTLY, and play the gliss loudly starting below and winding up above the E flat. Then release the pedal, etc.

Best,
Rick

The last not priority design on the 62 voice Korgs handle the drop outs better than the 126 voice Yamaha's..and the Roland 64 voice[G1000] handles this even better than the Korg..
--------------
my 9000 does'nt even come close to losing it's horsepower with it's 126 engine capabilities..can you tell me exactly which voices you were using..i cannot fathom a gutless 64 voicer outperforming a 126'er..!

OK, I finally got the $600 Yammie at GC to exhibit some note droppng. It was just a gliss up from middle C with flute and strings layered together. After exactly 16 notes a bunch of the lower ones dropped out. When I started with a low E flat and D flat -- held with pedal -- neither of those two dropped out. So it gives precedence to notes separated from others by time and/or pitch and/or harmonics. Not sure which. Didn't think I could stand there forever playing glisses, thinking up tests.

When I switched to piano/strings -- which is what I tried the other day -- polyphony became 32 notes. Which I spose is cause they have a piano/string "voice" in ROM. So it switched to that without informing yours truly. VERY hard to hear anything drop out with piano/strings.

I also learned that Kurzweils generally have four oscillators per voice, and when they say "32 note polyphony" they really mean 192 oscillators max. Kurzweil should make it very clear in all their specs that their polyphony means 2 to 4 times as many voices as a Yammie or Privia. (Roland? Korg?) How many people cept me are crazy enough to dig this out before making a purchase?

So the difference between the home market and the pro/stage market appears to be about 32 notes of polyphony. Plus you get an extra glob of sounds/features for a buck in the home market because they sell so many more home machines, and software and memory are cheap vs. oscillators.

Thanks for your patience. It's all coming together now. I think.

Rick

Mr9000...simply mathematics to start..The 9000 uses mostly 4 voices[polyphony]to produce a single tone, as the G1000 uses mostly a single voice to make a tone[sometimes two]...So on a common tone, the Roland [G1000]will have a 64 voice[polyphony] tone, where the same sound on the 9000 will be reduced to 32...

The same with the Korg , they use 1 and 2 voices to make a tone..

But more important than this....is allocation..

I played the 9000 and absolutely run out of voices...I am sure other 9000 owners will attest to this..Esch,Pesce,Boyd[I am being formal]..].

are poly counters always 100% accurate then?..i'm using lots with my accmp and everthng else and it is always in the area of 115-ish?

I would always dropped notes when I had 64 polyphony. With 128 - I've never had a problem, even when I use multiple layered voices with full styles.

are you running a Tyros2 Chony?

about polyphony

it's not 2 tone layered or 4 tone layered roland /yams

There are always patches in all synths that use only 1 waveform.

eg.
64 voice piano in roland xp series
some string/analog progs in Korg 01 and x series,even 16 poly M1 has programs that use 1 waveform

polyphony dropouts started with multi layering of 2 tones in program/patch modes mainly started in mid 90's(I'm not talking about combi/performance modes)where most synth players demanded for thick/layered patches to solo/or sustain chords -eg piano/strings

In a workstaion/smf playback world,32 poly nonGM 01W(though some progarams use 2 waves) or 24 poly GM sound canvas(SC 3x series)had never had much complaints in seq play back.(most midifiles were written for SC series as standard,Roland was the first to introduce the commercially sucessful GM modules in 92 ,and ofcourse GM arrangers E15,70,and synths -jv30(all with sound canvas board).

But in a live band situation you may not need 64 notes simutaneously,so layering up (eg-2 to 4 tones)is OK since players demand for sound of fuller stacked modules.

Polyphony is not a lie.it just happen to shift from early-mid 90's straight forward 1 wave-1patch )to multi wave layered patches after late 90's , because KB makers trying to make fuller individual patches(to impress the potential buyer compared to competitor),so ofcourse they do sometimes suffer from dropouts in seq/smf situation .

synth wise(not talking about early roland gm/gs arrangers)
Roland synths - 4 tone patchs (some of them are 2 or 3)
Korgs - 2 tone patches .

But arrangers are different.They need polyphony,KB maker/sound-style programmer must make sure that there should be no drop outs with regular playing,occational 2 patch layering, etc.Dynamic allocation helps but styles must not use too many layered voices or if they want to do so they should increase the polyphony accordingly.



[This message has been edited by jamman (edited 04-15-2006).]

No, the styles should be able to use good tones, but the modern keyboards, especially the expensive high-end ones, should allow enough polyphony to support that.

You mentioned the early Rolands (E35/70) - they were based on the 28-voice poly Sound Canvas, because this was the mainstay of the sound reroduction (Rompler) technology in 1991. There was practically nothing on the market which would support more polyphony than that. Today, with software based instruments, manufacturers can put in a faster processor, and increase polyphony just by modifying one memory location in the program; instead they are often trying to hamstring their sound engines from top of the line "professional" boards before stuffing them into an arranger.

The polyphony issue is very real indeed. While some people may experience it sometimes, and others may not experience it at all, whenever it happens to me, it occurs in the most inappropriate times - when I have a crowd dancing in front of me, and I am trying to adjust my playing to stay in sync with their energy.

Chances are, you will not notice much impact of the dropped notes if you are playing a single instrument, but when your ear relies on hearing the arranger style behind the solo voices, having notes drop out is like losing an auditory que, which is both displeasing to the audience and disconcerting to the player.

Regards,
Alex

what did I say

I said

Originally posted by jamman:
if they want to do so they should increase the polyphony accordingly.