Question: custom voice with looped samples

[BogdanH]

Question for those who created custom voices by using looped wav samples.
I know how to make custom normal voices from "normal" wav samples. I'm telling that so you know that I at least have basic knowledge 

Now I'm trying to make a voice where looped wav samples are used. That is, by using normal wav samples, I defined loop inside them. Resulting voice has span of two octaves and for each key separate sample is used (means, no pitch shifting).
If I import such voice in keyboard, everything works and sounds as expected... except if voice is used in the style. Let me try to explain:
The problem happens at changing chord. The first note (or first notes that start at about the same time) at the beginning of notes sequence, doesn't sound properly. For example: I have four quarter notes in 4/4 measure -in this case 1st of those four notes doesn't sound right in 90% of cases when chord changes. However, next time (when bar repeats), that 1st note sounds as it should. Btw. it doesn't matter if first note starts exactly at bar start or 50 ticks later -the problem is the same.
About doesn't "sound right"... It's not that some wrong note is being used -it's some distorted (random pitch) sound that doesn't even exist in custom voice.

At first I assumed I have some style setting wrong (those that influence "on chord change" behavior) , but that wasn't the case. Because I replaced my looped voice with some built-in voice, and in that case this problem doesn't happen. I also tried with custom piano voice that I created and everything works fine as well. I also tried with some 3rd party custom (looped) voice and again, no problem.

From what I said so far, I assume there's something wrong with my custom voice. Maybe there are certain rules for such voices? For example, minimal or maximal loop length? Or minimal/maximal time before loop should start? Or maybe I don't see something obvious?

Yes, I could try to figure out, but making looped samples, importing them into YEM and finally import into keyboard... all that is very time consuming.

Bogdan

[BogdanH]

Because there's no response, I assume not many are creating custom voices. Still, for those who might try in the future, here's what I additionally tried and my conclusion.

By studying some custom looped voices from Yamaha expansion packs (that do work correctly when used in style), I realized, that loop length is always extremely short: around 0.2 sec or less. What does that tell us? Basically it means, that the voice samples are artificially generated. That is, not a real instrument or vocal has been recorded. Human vocal (when singing certain tone), has much longer vibrato than 0.2 sec (depends on pitch)... and loop length must be set according to vibrato length -only in this case it will sound good. Hence, no surprise why all vocal voices that serve for padding (Mhm, Aah, etc.) sound quite badly.

Anyway, I reduced total length of the sample to 2.5sec (that's the length that most looped custom voices have) and shortened loop length to 0.2sec. Yes, it doesn't sound good, but I was curious if length is causing the problem that I described in my previous post. Btw. in YEM, I also used the same settings as used in 3rd party looped voice.
And result is: the problem remains.

I checked for few voices in Yamaha packs, to see what YEM version was used. But that info ("creator") is empty and so I checked in another 3rd party voice and there YEM version 2.5.2 was used. I couldn't find that version on internet, so I used te oldest YEM version that's compatible with SX700/900 (v2.5.3) and created voice in it again. And.. problem remains.

I just don't have any further ideas. It seems like YEM doesn't create looped voices the same way as that was the case with version 2.5.2 and I have no clue where the difference might be.
Anyway, I'm done with that.. for now.

Bogdan

[Misu]

Salut Bogdan,

Try the respective voice in another style or write a new style (track) directly in the style creator using it.
Something similar happened to me in styles coming from Cubase and only when changing the chord necessary to follow the melodic line.

Regards!

[BogdanH]

hi Misu,
For testing purposes I used a simple style that I created in Style Creator. It has relative slow tempo and I only used few (longer) notes, so I could easier follow chord changes and resulting sound.
Anyway, thank you for idea 

And the problem is...
Before I continue.. I don't give up easily and so I have found relative old YEM v2.3.2 and that didn't solve the problem either.
To keep story short: YEM version doesn't matter, loop position and loop length inside wav sample doesn't matter -I was just wasting time.

The thing is, none of looped voices work properly when used in style! -yes, you read that right.

When I was searching for solution, I only used padding voices (Choir Aah and similar). And because they're simple, the issue isn't that obvious -one needs to pay attention and focus to hear the problem. Anyway, when I noticed that the same problem exist with built-in padding voices, I tried to use some built-in saxophone voice instead. And in this case, it happens exactly the same as with voice that I created.

My final conclusion
Looped voices don't work properly in styles on PSR-SX700 (and probably on SX900). Actually I'm quite sure that this is a common problem for all PSR keyboards. No idea about Genos, though.

PS: After some further testing... This problem actually exist with all voices (not only with looped ones). But because non looped voices (i.e. piano) usually have very short decay, that issue isn't that much noticeable.
What a shame...

Bogdan