Broken guitar voices on PSR-SX700/SX900

[BogdanH]

I started learning some guitar solo music, where I decided for SemiAcoustic S.Art guitar voice, as it was close enough to original sound (at default settings). So you can replicate the problem, here are first few keys on keyboard of that music:
G3-F4-E4-E3-D4-C4

I immediately realized, that E3 key is quieter than the rest. At first I thought it's an S.Art "feature" that causes that, but no matter how hard I hit E3 key, it remain quieter. Then I decided to turn off Initial Touch, but that didn't solve the issue. After that, I turned off Compressor/EQ, but result was the same: E3 was still quieter in my ears. Finally, to exclude my hearing, I recorded a wav file (on USB stick) of all C3..B3 keys and analyzed levels with audio editor on PC.
Result confirmed my observing: there's at least 3dB difference! That is, keys C3-D3-E3 are 3dB quieter than F3-G3-A3-B3 keys (3dB drop is between E3 and F3).

Someone might say 3dB is a "small number".. well, it's big enough to be noticed even by an amateur musician (like me). Actually, the difference makes the voice unusable for solo playing.

Affected voices (all have the same icon):
SemiAcoustic S.Art
JazzGuitarClean S.Art
JazzArtistGuitar S.Art
JazzGuitarAmp S.Art
JazzGuitarSmooth SW S.Art
SlideJazzGuitar (not S.Art)
-interesting enough, SolidJazzGuitar S.Art voice is not affected.

Guitar voices which have "classic" guitar icon seem to be ok: that is, no 3dB drop at E3. I didn't check various electric guitars and Legacy guitars.

Bogdan

[travlin-easy]

Try performing a minor reset by holding down the rightmost white while turning on the keyboard,

Gary

[DerekA]

If you've used the same velocity, tt's almost certainly due to a different underlying wave sample being used over those two note ranges.

Bogdan, don't take this to heart, but the fact nobody else has complained about this means it is not 'unusable'. You just don't like it, and that's fine. Hit those keys harder to compensate now you know about it and are looking for it.

[EileenL]

Just tried these on my SX900 and all work as they should. No difference in volume.

[pjd]

Here is the voice meta-data from the SlideJazzGuitar UVF file. This voice has four elements. Each element has its own waveforms (multi-sample):

El#  Name     VelLo  VelHi  Wave#
---  -------  -----  -----  -----
  0  JazzGit     1     56    5679
  1  JazzGit    57     88    5680
  2  JazzGit    89    108    5581
  3  JazzGit   109    127    5682

Individual waveforms (samples) are assigned to MIDI notes in the following way:

Element key banks layout:
   NoteLo  NoteHi
   ------  ------
      C-2  E1
      F1   F#1
      G1   G#1
      A1
      A#1  B1
      C2   C#2
      D2   D2
      E2   D#2
      F2   F#2
      G2   G2
      G#2  A2
      A#2  C3
      C#3  D#3
      E3   F3
      F#3  G3
      G#3  A3
      A#3  A#3
      B3   C4
      C#4  D4
      D#4  E4
      F4   F#4
      G4   A4
      A#4  C#5
      D5   G8

Usually, Yamaha compensate for variations in waveform volume at the waveform level. This info is not reflected in the voice meta-data (UVF).

BTW, the first and last "samples" are stretched quite far below and above the natural range of the instrument. Players should stick to the range E1 to E5 if they want realism.

All the best -- pj

[BogdanH]

..
Bogdan, don't take this to heart, but the fact nobody else has complained about this means it is not 'unusable'. You just don't like it, and that's fine. Hit those keys harder to compensate now you know about it and are looking for it.

No worries, I don't take  these things personally. Doesn't matter if nobody complained so far.. I was reporting my finding (which is easy to replicate) and I stand to that. I said unusable for solo playing, because one can't easily notice that when strumming chords.
As I said, hitting key harder doesn't help, because even at full hit, the tone E3 doesn't reach the same level as F3 (because F3 is already relative loud). That was the reason why I disabled Initial Touch (because I was thinking I don't hit key hard/fast enough). Yes, I just don't like that E3 is 3dB quieter than F3 next to it.

Here is the voice meta-data from the SlideJazzGuitar UVF file....

I must admit, that I only have a vague idea of waveform numbers you show and absolutely no clue how and where to get this data for particular voice. Those who know that, could maybe compare the structure of two voices of interest.

It's interesting to me, that no one (so far) can't confirm what I am describing.. To make it sure: when I say E3 and F3 note, I mean the 3rd octave on keyboard -that's not 3rd octave of instrument (because when guitar voice is selected, it's octave is set to -1 by default).
Btw. the issue I'm describing is also visible in Compressor Output "LED" bar (in case you can't notice loudness difference of 3dB): turn off Touch Sensitivity, play keys in 3rd octave (with pause between each key) and observe the bar. You will see that E3 is quieter than F3.

Ok, I'll rest the topic at that.. unless some new finding appears.

Thank you all for commenting,
Bogdan

[Amwilburn]

I can't say I've been able to recreate it, but I think I know what you're talking about; when I go down, one starting with one of the E's it's a lot less bright... probably  because we're going from unwound to wound strings (wound G is generally preferred by jazz players) and when you play a higher fret, it gets a *lot* more muted and less 'percussive' than an unwound G. Play a scale on an electric guitar with a wound 3rd (G) string and you'll see what I mean.

I did go back and check to see if this is the same Semi Acoustic from the T5 (it is not) and to see if maybe you were using pull-off/hammer on notes (which are intentionally duller) but based on the notes you said, also very unlikely. Most likely it's simply the transition to unwound strings as you go up... they (unwound strings) tend to be more percussive.

Mark

[BogdanH]

hi Mark,
Yes, besides being noticeable quieter, E3 is also "less bright"... Anyway, I see you at least noticed there's a difference between E3 and F3.
I'm not a guitar player, but from what I've read, wounded strings are used because thick plain (unwounded) string would be too stiff. By wounding thinner string, string stays flexible and resonates better -sounds "brighter" than if unwounded thick string would be used. That's also the reason why top three strings are always wounded (G string is personal choice).
Which makes me thinking... the way "affected" voices sound, I can imagine that's how guitar would sound if all strings would be unwounded -but as said, I'm no guitar player. I guess we will never know the answer 
Anyway, I decided to use unaffected SolidJazzGuitar S.Art voice. After a lot of tuning, now it sounds the way I want.

Thank you for commenting
Regards,
Bogdan

[BogdanH]

Here's my additional finding...
First what I did:
I copied original SolidJazzGuitar S.Art voice (which is not "affected") on USB stick and opened the file on PC with hex editor. There I replaced only MSB,LSB & PC with values for SemiAcoustic ("affected") voice. That is, SemiAcoustic voice is used with SolidJazzGuitar settings (if I understand this stuff correctly). Then I loaded modified voice back into keyboard.

The result was... Obviously, it sounded differently than either of those two voices (which I expected). But important is, there was no loudness issue anymore! My conclusion was, it's voice settings that cause the problem (not voice sample). But which setting is that? I narrowed it to Effects -some of them just make that "drop in loudness". Not only there are a lot of effects, each effect also has a lot of settings -and because of that I decided to stop my "investigation".

That's it. I just though it's fair to report my finding, so the topic doesn't hang in the air.

Bogdan

[pjd]

I copied original SolidJazzGuitar S.Art voice (which is not "affected") on USB stick and opened the file on PC with hex editor. There I replaced only MSB,LSB & PC with values for SemiAcoustic ("affected") voice. That is, SemiAcoustic voice is used with SolidJazzGuitar settings (if I understand this stuff correctly). Then I loaded modified voice back into keyboard.

Bogdan

Hello Bogdan --

I'm assuming that you saved the voice to a file with the ".sar" extension. The kind of file is more generically known as a Voice or "VCE" file. "VCE" is the original extension for Yamaha voices and "sar" is the extension for Super Articulation voices.

VCE files are actually Standard MIDI Files (SMF) in disguise. If you change the "sar" extension to "mid", you should be able to open the "mid" file in a DAW (editor) like Cubase, Sonar, etc. There you will find the MIDI messages (including System Exclusive) that set the voice parameters and effects. No hex editor needed.

The UVF file type that I referenced is an internal file (format) used by Yamaha Expansion Manager (YEM). It takes a little digging to find the UVF files for legacy voices. Yamaha do not provide UVF files for the latest and greatest voices. UVF is XML with pre-defined tags. The UVF (Universal Voice File?) contains deep meta-data about a voice.

VCE is what users see. UVF is what YEM sees and is an abstraction of the hardware parameters.

This is probably TMI except for nerds. 

Hope this helps -- pj

[Enildo]

I immediately realized, that E3 key is quieter than the rest.

Not only the E3 but also the D3 and D#3

[rikkisbears]

Hi Bogdan,
I noticed a difference too. Under normal circumstances, I wouldn't have, but since you pointed it out, I tried it.
Played it on sx900 as well as via PA5x weighted keys. Same result.

[BogdanH]

...
VCE files are actually Standard MIDI Files (SMF) in disguise. If you change the "sar" extension to "mid", you should be able to open the "mid" file in a DAW (editor) like Cubase, Sonar, etc. There you will find the MIDI messages (including System Exclusive) that set the voice parameters and effects. No hex editor needed.
...

hi pj,
I know voice (description) files are actually midi files. And by knowing MSB, LSB & PC are near the beginning, it was much faster for me to change the values with hex editor.
Thank you for further explanation.

@rikkibears
Thank you for confirming. And you're right: it can only be noticed if you know what to look for, or you need to play certain notes sequence at certain loudness (touch sensitivity).
Here I'm playing the song where that happened to me: River of tears  - if you care to listen.
What I find interesting is, you say it the same on Pa5X?.. means, this "characteristic" might be done intentionally... and that those voices settings aren't really "broken".
Well, whatever the reason, now I know that if needed, I can "correct" that in Effects settings.

Bogdan

[rikkisbears]

@rikkibears

What I find interesting is, you say it the same on Pa5X?.. means, this "characteristic" might be done intentionally... and that those voices settings aren't really "broken".
Well, whatever the reason, now I know that if needed, I can "correct" that in Effects settings.

Bogdan
[/quote]


Hi Bogdan,

No, sorry , didn't explain it well. What I meant was I played those notes on the sx900 via my PA5x.
I have my PA5x and sx900 connected via midi. I'm using PA5x as controller. I no longer use the sx900 keys for playing .Only reason I tested with both sx900 keys and PA5x keys, was just in case there was something odd about those particular  keys. ( though, not likely).

[BogdanH]

Thank you for clarification!
Bogdan

[lmederos]

Someone might say 3dB is a "small number".. well, it's big enough to be noticed even by an amateur musician (like me). Actually, the difference makes the voice unusable for solo playing.

Every 3db means power is cut in half.

[BogdanH]

That's correct.
However, that doesn't mean -3dB is half the loudness. We percieve "half the loudnes" at 10dB change and so, for human ear, -3dB is only "a little" decrease of loudness.