CDP-S350 polyphony test

[Francisco Ruiz]

Now that the CDP-S360, having 128-note polyphony has been announced and will soon be available, I had second thoughts about the CDP-S350 that I had just purchased on sale, and which supposedly only has 64-note polyphony (and no Bluetooth MIDI). Should I return it while there is still time, and wait for the S360 to arrive?

 

Both keyboards have the same AiX sound engine, and everything else (except the Bluetooth) appear to be the same, so I made a test to see exactly how bad was the 64-note limit. Could I live with it, or would it feel unbearable at some point? This is the test:

 

1. Leave the Stage Piano sound on U1, and switch on Layers.

2. A Strings sound will load on U2. Change this to something with zero sustain volume, like a harp or a bass.

3. Step on the sustain pedal and hit one of the lower keys hard. I found that F works pretty good for this test. Then, without releasing the pedal do a soft glissando on the white keys up from about one octave above that key (so the previous single note still sounds distinctively). I began with a C.

4. Much to my surprise, the initial note was still heard, so I added another soft glissando on the black keys. Notice where the initial note ceases to be heard.

 

This may take a few tries because the initial note decays over time and the additional notes from the glissando may mask it (this is why you must stroke them just hard enough so they register). In my test, I covered five octaves of white keys plus another 15 or so black keys before the initial note seemed to drop out.

 

Since I had two layers going, each key was using two notes of polyphony, and here's how many notes were playing before the next note took the initial note's place: 2 x (1 + 7x5 + 15) = 102

 

This is certainly higher than the 64 that the S350 was supposed to max at.  I suspect it might actually be as high as 128, only my ears could not detect the initial note too soft to be heard but still there.

 

So here's a theory that you guys might want to discuss. Maybe the S350 and the S360 are actually the same, except for the Bluetooth. Maybe the only difference is in the firmware, and the last batch of S350 keyboards got the polyphony upgrade for free. Maybe I got the Bluetooth upgrade as well and I don't know it.

 

Do the test with yours and see if you get a similar result. Don't forget to report how old your instrument is, or what firmware version it shows at the end of the "Other" menu. Mine is 01.05

 

[Brad Saucier]

Or, maybe Casio's specs are true for all CDP-S350 units and it is 64 note polyphonic, but the algorithms Casio uses to prioritize notes is good enough that you'll never notice notes dropping out?  I tend to believe the published specs.  

[Jokeyman123]

And firmware would never be able to change the polyphony count without a concurrent change in the entire hardware design/architecture.

 

Polyphony is dependent upon the entire design of a keyboard-the number of oscillator voices that are available at any given time is determined largely by the hardware, and no amount of software modifications would be able to upgrade the polyphony if the capability is not designed into the machine. Why companies have taken so long to up polyphony counts from 32 to 64 to 128 to 256 and beyond, and it costs money to up the polyphony count-to create more circuitry and more powerful IC's to create distinct voices. Notice too-these numbers coincide with computer logic, and are partially based upon computer logic circuitry-32/64/128/256 bits-mathematically this follows the scheme for binary mathematics and the width of a data buss. Note priority algorithms can fool you into thinking there are more voices available than there actually are based on the hardware design, as Brad has described.

[Francisco Ruiz]

Both good points, Brad and Jokeyman123. I've heard about note prioritization, too. So now my question is: what does that consist of? How did the processor in my test know that it had to keep the first note going, perhaps at the expense of dropping notes that had started later? Are they perhaps ranked by velocity, by instantaneous volume?

 

Casio is not saying the S360 (and the S160, for that matter) has a new AiX processor. That really would cost development money. So I suspect the processor is still the same of the S350, and the change to up the polyphony count has been made in software. I own a lot of synths for my iPad, all with very different polyphony counts, and yet the processor is always the same. Many of those synths even allow you to set the polyphony within the software.

 

I believe the AiX engine came after the AiR engine, which has 192 of polyphony. I think it is entirely possible that Casio built a larger capability into the AiX but did not announce it publicly to avoid sales competition with their older, more expensive instruments. Then the CDP line got dinged in a lot of reviews for its relatively low polyphony, and the change to the S360 seeks to correct this.

[Francisco Ruiz]

I just made another couple tests, in case the glissando test did not really push the polyphony to its limit. Again, the test subject is a brand-new CDP-S350 with the 01.05 firmware.

 

** 1st extra test involved adding a Split layer to the U1 and U2 layers (stage piano plus ukulele), with an organ sound so it would not fade so long as the sustain pedal is depressed. With this trick the glissando can be as loud as you want because those notes will fade eventually, leaving the initial single note of  the organ voice intact, if it's still there.

 

**2nd extra test. Instead of a glissando, I pressed all the keys in the keyboard at the same time (except the initial A1) with a 4-foot construction ruler.

 

Results:

Contrary to expectations, the initial A1 note (I also tried some other low pitches for the initial note, both loud and soft) *never* cut out, even after pounding the other keys liberally to try and make them displace that note. Even if other notes were played before the A1 (or whatever it was). I just couldn't get any note to cut out because of reaching the polyphony limit. Maybe it could be reached by recording several tracks and making them play simultaneously, but I haven't tried.

 

So, unless some really amazing trick is at work, now I'm measuring a polyphony of at least 2x88 = 176 for this CDP-S350. It could be even higher, but I haven't thought of a way to test it. Any ideas?

[Joe Muscara]

Not that I mind you testing, and I'm always interested in the results of a good test, but how much does it matter? Are you playing or creating compositions that use that many notes? Are you driving the CDP-S350 with a DAW pushing a ton of notes into it? (Which would probably be the easiest way to test polyphony, BTW.) 

 

If Casio's algorithm is really good, you may never be able to tell what the polyphony really is, like @Brad Sauciersaid. But if testing it is a fun project for you, go for it. I would just suggest that you don't drive yourself and anyone near you crazy trying to figure this out.  

[vitoruch]

I've heard somewhere that new models had a better, more longing decay..wouldn't it influence in polyphony?

One more thing - Heard that internal Speakers, although same 8 Watts, were a little better sounding - Is it true?

 

It's incredible how, after several weeks from launching (and available for buying in a few countries, at least) , there is no review in whole internet yet! 

[Francisco Ruiz]

Good points, all of them. I guess what my tests show is that it really doesn't matter what the numbers shown in a specs card are. What matters is performance in the real world. The CDP 350 and CDP S100 got dinged in a lot of reviews for their allegedly low polyphony count of only 64 notes. But if in a practical test they behave as if the number is actually 176, what does it matter what the official number is?

 

In other words, don't just rush out to buy the CDP 360 because it is supposed to be "better". You'd be hard pressed to find a situation where you can tell the difference. But I'm still curious as to whether I'm getting this result because it really doesn't make any difference, and it never did, or because I have a very new machine with a very new firmware where the "problem" has been solved quietly.