I have a problem with lag on my S70XS – just wondering if anyone can help me.

I am using a performance that has 4 voices - it’s a mix of piano, strings and 2 pads

Two of the voices are edited from the original and all the voices apart from piano have restricted key ranges. Not sure if that information is at all relevant though!

The problem is that - sometimes - when I am playing a lot of notes together as a chord, not all the notes sound at once! There is a fraction of a second delay in some notes and it can sound like an ever so slightly broken chord.

I thought this was me not hitting the notes properly…..until my son played the board the other day and said “wow, what’s going on here?”

So – what is going on?
If I disable one of the voices, there is no problem.
Is the processor being overloaded? Surely this cannot be the case, it’s one of the things that Yamaha would have exhaustively tested for. Yet that is exactly what it looks like.

Any tips for what to do or try (short of playing with only 3 voices…)?

I did find this here which looks to be very very similar to my experience

http://forums.musicplayer.com/ubbthreads.php/topics/2408053/Re_Weird_S90XS_problem_Piano_G

Many thanks!

It seems you’re describing a polyphony issue.

When polyphony is sufficient for what’s being played, previously struck notes whose envelopes haven’t fully run, along with new notes, will play without consequence.

A held note continues to use polyphony until the note is released. Note holding can be either due to keeping a key depressed, or use of a sustain pedal.

When you run out of polyphony, what occurs is “note stealing” - notes with low priority drop out, and are replaced with ones that were triggered somewhat later. “Note priority” is determined by an algorithm, which can vary from keyboard model to model and with manufacturer.

Polyphony usage can be calculated. Each Element in a Voice that’s active when a key is struck uses one note of polyphony if the Waveform is mono, two if stereo. Count all the Elements that are in play for a Voice, times two if stereo, add that to the count for any Voices that are layered, then multiply that by the number of notes being held, plus one for the next note (or more if a chord). If that number exceeds the rated polyphony, there can be audible results, more obvious the greater the polyphony limit is exceeded.

Thankyou for replying so quickly.
The max polyphony is stated as 128 notes...but from what you say, if a voice is 8 element and it’s in stereo - such as a piano - then one note is consuming 16 notes worth of polyphony - ie the real world polyphony of that voice is actually only 8 notes ( ie 8 x 16 = 128). Then add the other 3 voices...!
Maybe the polyphony consumption is not quite so drastic as that?

I’m away just now and will have to experiment with different voices when I return - I will post again with my findings!

wp54 - 01 January 2017 06:04 AM

Thankyou for replying so quickly.

You’re welcome.

 

wp54 -

The max polyphony is stated as 128 notes...but from what you say, if a voice is 8 element and it’s in stereo - such as a piano - then one note is consuming 16 notes worth of polyphony - ie the real world polyphony of that voice is actually only 8 notes ( ie 8 x 16 = 128). Then add the other 3 voices...!
Maybe the polyphony consumption is not quite so drastic as that?

Indeed, it’s not quite that drastic. It’s rare that all eight Elements of a Voice are active at once, even if they’re “on” - Parameters such as note range, velocity range, and XA Control determine whether a particular Element is actually playing at any one time. (There are some Voices that use all eight Elements simultaneously - certain organ sounds, for example.)

For factory Preset Voices you can get an idea of usage by looking at the “Polyphony/Note” column of the Voice List in the S90XS/S70XS Data List (http://download.yamaha.com/file/9637). Of course, for edited Voices you might have to make your own determination. Use of the Assignable Function buttons for XA Control might increase or decrease polyphony consumption.

 

wp54 -

I’m away just now and will have to experiment with different voices when I return - I will post again with my findings!

OK, do let us know what you find.

Hi - apologies this is late, however, I have an answer.

I created a performance with 4 pianos that each contain lots of elements - ie to make the worst possible polyphony situation so that it’s easy to see where the problem shows itself.

- If I play lots of different chunky chords (ie many notes) one after the other with the sustain pedal held down - then I would expect the most severe problem (with note play delay) as this consumes polyphony in the max way I can think of.  But - NO PROBLEM!!

- If I play those same chords but correctly pedal from one chord to another (ie as the new chord is played, lift the sustain pedal and put it down again) - NOW the problem of note delay appears bigtime.

So I don’t think it is entirely a polyphony issue - in that second instance, the polyphony demands are LESS but this is where the problem occurs

I do think it is a processing speed issue when so many input signals (ie combination of many elements + sustain pedal) are happening.

Any thoughts?

wp54 - 28 March 2017 05:43 PM

Hi - apologies this is late, however, I have an answer.

I created a performance with 4 pianos that each contain lots of elements - ie to make the worst possible polyphony situation so that it’s easy to see where the problem shows itself.

Even though piano Voices often use all eight Elements, polyphony usage is typically only 2 per note - multiple Elements are used to cover velocity and note ranges, but are not active simultaneously. See the Data List previously referred to.

 

wp54 -

- If I play lots of different chunky chords (ie many notes) one after the other with the sustain pedal held down - then I would expect the most severe problem (with note play delay) as this consumes polyphony in the max way I can think of.  But - NO PROBLEM!!

- If I play those same chords but correctly pedal from one chord to another (ie as the new chord is played, lift the sustain pedal and put it down again) - NOW the problem of note delay appears bigtime.

So I don’t think it is entirely a polyphony issue - in that second instance, the polyphony demands are LESS but this is where the problem occurs

I do think it is a processing speed issue when so many input signals (ie combination of many elements + sustain pedal) are happening.

Any thoughts?

MIDI can be subject to data overload, also sometimes referred to as MIDI “choke” or “clog”. Note-on and Note-off events can be part of data that can cause this, but a considerable amount of controller data can also contribute.

Before I get more deeply into the issue of possibly excessive data…
Is your S70XS connected via MIDI to anything else? If so, disconnect the cables and let us know if the problem persists.

Thankyou again!

Yes - this is without the S70XS being connected to anything else - just powered on and go.

I get your point about piano voices not using all the elements at once - nonetheless I have been able to make the note delay issue MUCH worse by making that multi-piano program!

It’s also worth adding that I have defined note ranges in my proper program for each of the voices except piano (ie the 3 strings/pads) -ie one pad only plays in the upper register, another only in the lower - so again here, it is unlikely to be a polyphony issue, more - I think - pointing to a data processing speed issue.

Having said all, I’m pretty sure there’s not much that I can do in my real life situation, except only use 3 voices in my program..... which rather defeats the point of having 4 slots, but at least we seem to understand more of what is going on.

One final thing I can think of - when I have a chance I will do a factory reset and reconstruct the piano test program - just in case there is a global setting I have changed that is causing this.

You could try a factory reset. However…

Synths, of course, don’t have the ability to simultaneously play an unlimited number of notes. Synth designers have to use algorithms that do Dynamic Voice Allocation, in order to minimize the effects of the limitations. For certain programming and playing styles, that can still result in some audible glitches.

So yes, you may have to adjust programming of your Performances to accommodate the limitations.

You might find the following threads interesting:
http://www.motifator.com/index.php/forum/viewthread/465327/
http://www.motifator.com/index.php/archive/viewthread/329451/
http://www.motifator.com/index.php/archive/viewthread/342691/

Thanks again – and the threads all make really interesting reading.
In case I have confused anyone in previous posts, I don’t know why I called Performances “Programs”!!

As you suspected, a Factory Reset didn’t improve matters, so “it is what it is” and I will just have to work with it.

Having said that, it’s nevertheless interesting to pursue a little further.

Putting everything that I’ve done and read together, I can conclude that what (and I think others) I am experiencing is most definitely NOT a polyphony issue, but IS some kind of data speed processing issue.
However it is true that using a performance that has voices with lots of elements is the vital ingredient that makes the problem appear.

So why do I think not simply polyphony?

Two things really
1/ Using my original Performance (Piano + 3 pads/strings), playing different chords and holding down the sustain pedal results in no problem, whereas pedalling as I transition from one chord to another results in significant note play delays. This second instance will be making less polyphony demands but more data processing demands.

2/ That original Performance defines note ranges for each of the strings/pads so that they don’t all play at once. And the piano – as has been pointed out – will only play a couple of elements at any one time (actually 4 since stereo). So really the 128 element polyphony limit is, by my estimation, far from being reached.

Well there it is, I will now need to further adjust my Performance so that the problem goes away. One easy way - eliminate one voice....I will try reducing note ranges more first.

wp54 - 12 April 2017 06:45 AM

Thanks again – and the threads all make really interesting reading.

You’re welcome.

 

wp54 -

In case I have confused anyone in previous posts, I don’t know why I called Performances “Programs”!!

No problem. Performances are a subset of Programs - Voices, Masters, Songs, and Patterns are also Programs.

 

wp54 -

[...] Having said that, it’s nevertheless interesting to pursue a little further. [...]

I’d be interested in pursuing this a bit further as well. I’d like to try one of your Performances myself. If you’re willing to post an example, this should work…

1) See http://www.motifator.com/index.php/forum/viewthread/461620/ (be sure to read the Warning) and download my Init User Banks.X0A file. After backing up your own data to an “All” file, you should be able to load the .X0A to the S70XS, resulting in initialized User Banks.

2) Create a Performance that when played exhibits the delay problem you’ve described.

3) Save an “All” (.X2A) file, zip it, and attach it to a post in this thread.

Since my XF can load an S70XS “All” file, if you do the above I’ll see if the problem can be duplicated with the XF, and what might possibly be done to minimize it.

Ok will do!

I’ll have a chance to look at this again after Easter...watch this space!

Hi again – sorry for the delay.

I’ve done as you said, hopefully my “all” file is attached to this post.
I decided to create a performance using 4 S6 pianos – this is on USR4 no 19 – just to see what would happen. This is the worst example! It actually has problems even if only 3 of the voices are turned on.
Even with all 4 voices turned on, if different dense (many notes) chords are played in quick succession, if the sustain pedal is held down, there is no issue with note delay.
It’s only if the sustain pedal is used to pedal one chord to another that the note delay arises. 
So from this alone, all I can think of is that the problem is processing power, not polyphony.

As an example of the real performance that I spotted the problem on, try USR 4 No 14 (ignore that it’s called VP soft…!)

Then - I downloaded and loaded the .xoa file – and as you say it wipes the slate completely clean.
I created the same 4 voice piano performance and… same problem.

So – I just think we have found the limit of the board, only it’s not the limit we thought it was!

Just for interest also -Since then… I have acquired a CP5 – it also has the 4 layer facility – not a hint of a problem here.

Sorry, file was not attached. Hopefully now....

4 JUN 2017 S70XS.zip  (File Size: 296KB - Downloads: 571)