PSR e473 - "incoherent" sound of pitches in some voices. Question to e473 owners

[ancanar]

I made an interesting discovery about some voices in psr e473. First it occured to me on voice 002 (grand piano) that E pitch in second octave sounds a bit louder than adjacent pitches (of course with dynamics set to fixed). This did not occur on Live grand piano (001) but occured on some other vocices as well.

Later I did an experiment. I was playing adjacent keys with one with pitch bend wheel on full tilt to make them sound exactly the same pitch. In some voices, like Live Grand Piano, the pitches sounded (as it was expected) exactly the same but in others (like regular Grand Piano) there were significant difference in sound colour (although the pitch was exactly the same - checked with KORG eletronic chromatic tuner) almost like some kind of different equalization was used on different pitches. I found it occuring not only with afromentioned E pitch but with several other pitches all over the keyboard range, and this  was the cause of me percievieng the loudness differently, I believe.

Can this thing be a result of sloppines when recording samples for certain voices? TBH from a company like Yamaha I was expecting more precision... or maybe it is something wrong with my keyboard? (although i doubt it). I am curious if other e473 owners can repeat my test and will get same results. I will later also add videos of me doing said test (away from home right now).

Best regards and waiting for Your thoughts.

[BogdanH]

Now that you've mentioned that... I have PSR-SX700 and quite a while ago, I had the same experience on my keyboard (if I understood you correctly). I can't remember what piano voice I was using, but I can clearly remember how it sounded.. it was like particular key was slightly out of tune. That is, for example, you press B key and then C key.. but C key doesn't have the pitch (or "sound", if you wish) you expected to hear after pressing B key. I even disabled touch response, just to make sure I'm not hitting keys differently, but it wasn't that.
And same as you, I came to conclusion, that when making voices (by Yamaha), some voices just aren't that perfectly transposed.. and didn't bother any further about that issue. I can remember, that at that time, when I've found out about that, I kept hearing that "out of tune" note all the time LOL. But after a while, after using other voices, I entirely forgot about all that. And right now, I'm not even sure if it was an issue... The thing is, our hearing is not always the same  -not even during one day!

Wish you fun with E473
Bogdan

[ancanar]

Now that you've mentioned that... I have PSR-SX700 and quite a while ago, I had the same experience on my keyboard (if I understood you correctly). I can't remember what piano voice I was using, but I can clearly remember how it sounded.. it was like particular key was slightly out of tune. That is, for example, you press B key and then C key.. but C key doesn't have the pitch (or "sound", if you wish) you expected to hear after pressing B key. I even disabled touch response, just to make sure I'm not hitting keys differently, but it wasn't that.
And same as you, I came to conclusion, that when making voices (by Yamaha), some voices just aren't that perfectly transposed.. and didn't bother any further about that issue. I can remember, that at that time, when I've found out about that, I kept hearing that "out of tune" note all the time LOL. But after a while, after using other voices, I entirely forgot about all that. And right now, I'm not even sure if it was an issue... The thing is, our hearing is not always the same  -not even during one day!

Wish you fun with E473
Bogdan

Interesting. Like I mentioned I checked this with a tuner and the sounds were all on pitch. It is just like they have a bit diferent harmonic components. I believe the samples are recorded from real instruments so maybe they did it a bit clumsy? (however - no scuh issues on my old PSR-270) Still curious about expierences from other e476 users.

[SciNote]

I'd be curious to see if other people have this issue, as well.  I have an E433, not the E473, and never noticed it on my keyboard, but I also never really listened closely for that type of anomaly, either.

As I understand it, voices like piano are made up of several samples throughout the range of the keyboard, and not just one sample that is sped up or slowed down throughout the entire range.  Perhaps where you are hearing this change in timbre/volume could be the point where Yamaha went from one sample to another for that particular tone.

[pjd]

Hi --

Voice 1, Concert Grand Piano, is the featured piano voice and is the most recent piano voice added to the E series.

Voice 2, Stereo Grand Piano, is pretty old. I couldn't find a UVF file for this voice (MSB 0, LSB 112, PC 1). UVF files are XML format and contain voice meta-information.

I found a Genos/Tyros5 UVF file for this voice (MSB 0, LSB 115, PC 1). It is definitely a more advanced voice versus Stereo Grand Piano. Concert Grand Piano has three voice elements with 31 key banks per element. All key banks have the same layout across the full range of MIDI note values:

Key Bank  Key Low  Key High
--------  -------  --------
    0       C-2      C#0
    1       D0       E0
    2       F0       G0
    3       G#0      A#0
    4       B0       C#1
    5       D1       E1
    6       F1       G1
    7       G#1      A#1
    8       B1       C#2
    9       D2       E2
   10       F2       F#2
   11       G2       A2
   12       A#2      C#3
   13       D3       E3
   14       F3       G3
   15       G#3      A3
   16       A#3      B3
   17       C4       C#4
   18       D4       F4
   19       F#4      G4
   20       G#4      A#4
   21       B4       B4
   22       C5       C#5
   23       D5       E5
   24       F5       F#5
   25       G5       G5
   26       G#5      A#5
   27       B5       C#6
   28       D6       E6
   29       F6       G6
   30       G#6      G8

So, you can see that only a few individual waveforms (AKA "a sample") are assigned to a single key. The rest of the notes in a range (key bank) must be stretched (pitch bent) down or up to cover the key bank range.

You're right, you may be hearing pitch shifted notes as louder or softer. All of the tone harmonics are shifted along with the base, fundamental harmonic. Since piano tones are harmonically rich, the change may have a psycho-acoustic effect.

I wish I had a key map for the older Stereo Grand Piano. It probably has fewer elements and/or waveforms. Pianos need more memory than other instruments and Yamaha adds better pianos when more waveform memory is available in flash/ROM.

Hope this info helps your explorations -- pj

[BogdanH]

Thank you for explaining and confirming -so we know we're not hallucinating
And as Ancanar said above, keys have the right pitch (of course). It's just, when using certain voice, it can happen that two adjacent keys don't sound as expected when crossing key bank range.

Bogdan

[pjd]

Hi --

Thanks. I played around on MODX, too, trying the most basic grand piano patch of yore. I couldn't hear mismatches, but pitch bending up a semi- or whole-tone does mess with the harmonics.

-- pj

[HiHoSilver]

I've noticed the same sort of thing on many voices on my PSR-E463, too - that (for instance) playing a scale there will be sudden changes in the tonal character at certain spots on the keyboard.  It would have been nice if they'd smoothly blended the samples together (e.g. if I'm playing a note that's about halfway between two samples, it might be a roughly equal blend of the two samples), but oh well, perhaps that's too much to expect from a reasonably inexpensive consumer device.

[andyg]

FWIW, these are called 'sample breaks' and they've been around since instruments started using PCM samples. Some of the very first instruments had just monophonic samples, with maybe 3 samples across the keyboard. So a trumpet that sounded fine an octave above Middle C would play at half speed at Middle C and 'burp'! Memory was ferociously expensive back then, of course.

Things rapidly improved and the 'norm' became one sample covering three or four notes. All makes worked hard to disguise the sample breaks but you could still hear them. In fact, you still can, regardless of the instrument or where it sits in the range. I can hear them on my Roland Atelier organ, and that was a £25,000 instrument when new. Some breaks will be more obvious than others and where you go outside of an instrument's true range, it's common not to use any new samples, so you still get the speeding up / slowing down effect. It is possible to blend from one to another but it's tricky and no doubt costly to do.

It's one of the prime reasons that, back in the late 1970s, when Kawai were working on their first digital instruments, we decided against using samples for anything except drums. We'd sample, of course, but then use powerful computer analysis and create the sounds using real time additive synthesis. It was great 'fun' pointing out the advantage we then had - no sample breaks. We could ask a customer to play up and down an octave of a given sound and listen to our instruments versus rivals! When sampling 'came of age', we eventually switched.

If you want one sample (or multiple samples) per note on a piano, you probably have to go virtual. One of the virtual grand pianos on this PC is 11GB, and that's not a recent one. The start of each note is loaded into RAM and then the rest is streamed from disk. Same with virtual pipe organs.

There are some instruments with one/multiple samples per note, Allen organs, for example. Wersi and Bohm organs, being computer based (Windows in the case of Wersi) may well have this, but I haven't tried one in some time.