Howdy!

Has anyone seen this?

In Voice mode, go to patch Pre4:054 (Pre4,D,6), “3 Octave Strings”.

This is a 4-element string synth patch, that I’m going to assume uses stereo samples so thus it is effectively an 8-"voice" patch.
Thus, out of the 128 ‘voice’ poly this “should” use 8 of them for each active note.

What I am hearing though, is for example I leave the sustain pedal alone - and just play “C,D,C,D,C,D,C,D etc” say at 85 bpm quarter notes - so not too fast but not too slow.

What I am hearing is this - after about 8 notes, i will hear the tone generator not using all 4 elements for a keypress - sometimes even just one. This sound has 3 active octaves in a single keypress so it’s pretty obvious when this happens (it basically sounds much much thinner).

My math tells me that with 128 voices, assuming these 4 elements are stereo elements each keypress should be using 8 voices, is that I should be able to play 16 notes before any sort of ‘stealing’ goes on. But I have only played 8 of them, nevermind that they are not sustaining and that the issue is occurring with the NEW note, not an old decaying one…

Has anyone else run into this? Or at least can someone else please try this and see if you hear the same thing? All of the self tests pass, although admittedly there’s no self tests for the SWP51 that I could find except for the simple 1khz tone output…

Thank you!

hammondeggs - 11 April 2015 01:27 PM

[...]Has anyone else run into this? Or at least can someone else please try this and see if you hear the same thing? All of the self tests pass, although admittedly there’s no self tests for the SWP51 that I could find except for the simple 1khz tone output…

First, let me assure you that your XS6 is behaving “normally” - that is, no differently than my XS6 and XF6.

What you’re experiencing is certainly not a matter of available polyphony. Instead of repeatedly playing the same two notes, try doing a gliss/smear across all the keys, both white and black. I think you’ll find a great number of notes sustaining, to the point that detecting which might have been dropped becomes difficult.

What’s happening is characteristic of the particular note priority algorithm which Yamaha uses in the Motifs. Most of the time it’s not an issue, although under certain playing conditions it becomes evident.

This has been discussed before:
http://www.motifator.com/index.php/forum/viewthread/462823/
http://www.motifator.com/index.php/forum/viewthread/473540/

If you reduce the Attack Level in the AEG to 120 or below on all elements, the anomaly disappears.

Hrumph. I can understand how “old” notes would be replaced by new notes (even if the algorithm might be less than optimal..) but what I’m hearing is new notes not using all of the elements - for example playing the cdcdcd when I hear a “thinner” note it’s the one I’m playing - I hold the key and it’s still just one or two elements sounding - this is what you can reproduce on your xs/xf?

stoneb3: I will try changing the attack and see what happens, thanks!

I did try this though: with a new initialized voice, I assigned a stereo pwm string element to all 8 elements, with unique octaves for each - admittedly couldn’t get the same effect, but with 8 elements it might be harder to hear when one or two aren’t sounding…

Thanks!

Sounds like you’re describing phase
cancellation.

Anytime you repeatedly strike keys on a polyphonic voice,
you are stacking up envelopes on top of
one another. Especially if the sound has some
release (like strings) and doesnt cut off
abruptly (like an organ).

You may think you are triggering new notes
one after the other, but you are actually
“overlapping” previously triggered envelopes
with new key presses, and the previously
triggered envelopes havent finished their
cycle (because the release segment of
those envelopes is still playing).

Try reducing the release of the voice to
zero (you can do this with the release
knob on the frong panel) and see if the
behavior goes away.

I agree that phase cancellation is likely part of what’s being heard, because the “oscillators” aren’t locked in phase to each other. Try sustaining PRE2:005(A05) Fluty Pipe, and repeat just one note, for a distinct example.

However, but I’m not convinced it’s the entire explanation, based on previously noted behavior that has nothing to do with phase - see especially the second thread I linked to.

Under any circumstance, “it is what it is”. I don’t anticipate Yamaha changing anything, so it might be best to adapt playing technique to the situation.

Indexing and zeroing of the front panel control release knob does little and may make it worse. Prior to my previous post I reset the release values of the 4 elements to various lower positions with no success other than expectedly cutting the voice short, when I reached that point and below. I should mention that I was conservative in the release reduction as I went.
As I stated before when I lowered the attack levels of the 4 elements to 120 or below the matter disappeared, with no appreciable loss to the voice dynamics.  Don’t know why and I didn’t have a whole lot of time to dig into it. If you want rid of it, there you go.

With sustain engaged, the OS seems to allow only four “instances” of a note to be played simultaneously. While that can help with polyphony, I think there are other reasons for such behavior.

For every doubling of signal amplitude, the result is an increase of 6 dB. With a Voice that doesn’t decay, if a key is struck twice (and phase happens to align) the output will be 6 dB higher. If the same key is struck twice more (again assuming phase alignment) there would be another 6 dB increase, or a total of 12 dB over that of the key being struck just once. If there was no algorithm preventing it, multiple strikes of the same key could eventually cause overload.

Typically there isn’t phase alignment, so the amplitude increase wouldn’t normally be as much of a concern. However, unlimited key strikes with phase cancellation could eventually lead to waveform “soup”, sounding more like noise than music.

It appears that there are factors requiring more attention to detail in the OS than might be expected.

5pinDIN - 12 April 2015 01:04 PM

With sustain engaged, the OS seems to allow only four “instances” of a note to be played simultaneously. While that can help with polyphony, I think there are other reasons for such behavior.

For every doubling of signal amplitude, the result is an increase of 6 dB. With a Voice that doesn’t decay, if a key is struck twice (and phase happens to align) the output will be 6 dB higher. If the same key is struck twice more (again assuming phase alignment) there would be another 6 dB increase, or a total of 12 dB over that of the key being struck just once. If there was no algorithm preventing it, multiple strikes of the same key could eventually cause overload.

Typically there isn’t phase alignment, so the amplitude increase wouldn’t normally be as much of a concern. However, unlimited key strikes with phase cancellation could eventually lead to waveform “soup”, sounding more like noise than music.

It appears that there are factors requiring more attention to detail in the OS than might be expected.

Brilliant summation of the situation, 5pin.

To paraphrase the old joke:

PATIENT: “Doc, when I play more than four notes
successively it sounds bad.”
DOCTOR: “Then don’t do that”.

OK . I should point out that this occurs regardless of playing successive notes or not - I can hear it happening playing chords etc, but playing C-D-C-D repeatedly is an easy way to demonstrate the issue - the main part which concerned me was that without using any sustain pedal etc - the note that you held on to would indefinitely sound with just one or two elements - which to me seemed counter-intuitive to the traditional ‘note stealing’ where *previously* played notes would disappear as they were re-allocated to the newer notes, etc.

If you all are saying that you can reproduce this on your XS/XF’s, then I guess that’s the way it is then...!

Thank you!