NS3 polyphony

[baekgaard]

A good test would be to record an audio file of a sine wave as baekgaard did, starting with the sustain pedal pressed and play increasingly higher pitch notes. Then look at the spectrogram; I think it should be visible where a line in the spectrogram disappears when another appears (showing the note stealing effect). Sadly I have no NS3 to try

[MrTobbe]

A good test would be to record an audio file of a sine wave as baekgaard did, starting with the sustain pedal pressed and play increasingly higher pitch notes. Then look at the spectrogram; I think it should be visible where a line in the spectrogram disappears when another appears (showing the note stealing effect). Sadly I have no NS3 to try

Smart idea! I will try this tomorrow!

[alcools]

Stereo?
2 stereo samples = 4 polyphony per note.

Seriously ?
No, nobody's counting like that

That's exactly how it works on most workstations, such as the Motif series and others, that count "voices" by operators and that in case of stereo samples allocate 2 "voices" (one per channel/sample).

@Mr_-G-: I am afraid 34 is not a typo, because it is the same on the comparison charts; so if 34 is correct then, even assuming 2 voices being used in case of stereo samples, that would still not match lordy's numbers: any other NS3 here who can double check and test the same?

I learned something today so
I don't remember this kind of count on my precedent samplers: Yamaha A3000, Kord Triton LE, Nord Satge 2 or Nord Wave.
I will do some test this week end with my NS3.

[maxpiano]

A good test would be to record an audio file of a sine wave as baekgaard did, starting with the sustain pedal pressed and play increasingly higher pitch notes. Then look at the spectrogram; I think it should be visible where a line in the spectrogram disappears when another appears (showing the note stealing effect). Sadly I have no NS3 to try

Smart!

[lordy]

Forget the spectrogramme.
Play a big loud C, sustain pedal on, and play slowly cromatics notes. You ll see.
I think actual polyphony is enough, but nord have to be serious, if they say 34 notes, i must count 34 notes. It is not the case, by far. If nord say sample editor soon, november, a c2d inside, nord have to respect what they say, and stop to tell things which are not true. That is my opinion.

[baekgaard]

Forget the spectrogramme.

No, it's actually a good idea. It can also help to see if other tones drop out, so it's a more precise test than what you suggest.

Play a big loud C, sustain pedal on, and play slowly cromatics notes. You ll see.

That's what I did (almost -- I may have played an F in one of the higher octaves before starting the chromatic scale with the sustain pedal down).

I think actual polyphony is enough, but nord have to be serious, if they say 34 notes, i must count 34 notes. It is not the case, by far.

I typically get to 28 (panel B) to 30 (panel A) notes with this test for single-voice programs.

In my book that is close to 34 and not far from.

YMMV.

[PScooter63]

Forget the spectrogramme.
Play a big loud C, sustain pedal on, and play slowly cromatics notes. You ll see.

No, I think it is important to measure your claim precisely, and clarify it once and for all.
Also there are several different voice stealing/allocation algorithms, in the order played, or keeping the extremes, top priority, low priority, some overflow to MIDI, some analyse the phrase. It might be possible to detect in the spectrogram which one is implemented.

[PScooter63]

Play a big loud C, sustain pedal on, and play slowly cromatics notes.

Why slowly? (Why chromatics?)

You specifically mentioned samples in your initial post. Since it’s common knowledge that one can’t import custom samples at this time, what specific sample did you try, and what program?

(Could it have been a sample/program that does not loop, and comes to a natural end before you “slowly” reached the magic number?)

The only other thing I will say, because no one else will, is that you assume an awful lot when you accuse a company of lying. Lying is an intentional act, and you have not offered solid proof of Clavia’s intent (so far).

[MrTobbe]

Well, I did the frequency spectrum test in Audacity - I used the sine wave in the synth section, played a C scale of 34 notes while holding the sustain. I repeated it three times to be sure, but I got the same result in each test - there seems to be 30 notes of polyphony, now more, now less.

Here you can see the spectrum yourselves - the gap after the lowest frequency are most likely caused by notes cutting off, when adding more high notes. So here we also see that the lowest note has priority, but after that notes get cut off in the order they were played.

Spectrogram, sine waves, Nord Stage 3.png (147.68 KiB) Viewed 1859 times

[MrTobbe]

...and just to be sure - I have now done the same test, but with two identical panels of a pure sine wave, and layered panel A and B together. The maximum polyphony then is 15 notes.