How to: Improving the Pa2x "GrandPianoRX" sound

Rob Sherratt · Post by **Rob Sherratt** » Wed Apr 16, 2008 7:45 pm

Here are some "improvements" that can be made to what is arguably one of the best "Grand Piano" sounds on any keyboard on the market. Firstly, select the Grand Piano RX sound, and go into Sound mode on your Pa2x. Press Menu and go to the "Sound Basic" screen. You will see that 15 oscillators are used, and the priority of 5 means that Osc 11, 12, 13, 14, 15 (RX noises) will be the first to go if we run out of polyphony. If you look in the other two screens "OSC Basic" and "Vel/Key Zone", you'll see that sample assignments and usage are as follows:

Osc | MultiSample .| Velocity| Level |. Usage
1 .. | GrandPiano_L .| 104-127 |. 127 .| L speaker piano note, played ff
2 .. | GrandPiano_R .| 104-127 |. 127 .| R speaker piano note, played ff
3 .. | GrandPiano_L .| 089-103 |. 122 .| L speaker piano note, played f
4 .. | GrandPiano_R .| 089-103 |. 122 .| R speaker piano note, played f
5 .. | GrandPiano_L .| 045-088 |. 115 .| L speaker piano note, played mf
6 .. | GrandPiano_R .| 045-088 |. 115 .| R speaker piano note, played mf
7 .. | GrandPiano_L .| 002-044 |. 115 .| L speaker piano note, played pp
8 .. | GrandPiano_R .| 002-044 |. 115 .| R speaker piano note, played pp
9 .. | Tailpiece_L ......| 002-127 |. 127 .| L speaker piano resonance if sustain/damper is on
10 .| Tailpiece_R ......| 002-127 |. 127 .| R speaker piano resonance if sustain/damper is on
11 .| FX Key Off_L ...| 002-127 |. 005 .| Percussion noise when you release a note
12 .| FX Pedal On_L | 002-127 |. 060 .| L percussion noise when sustain/damper is depressed
13 .| FX Pedal On_R | 002-127 |. 060 .| R percussion noise when sustain/damper is depressed
14 .| FX Pedal On_L | 002-127 |. 060 .| L percussion noise when sustain/damper is released
15 .| FX Pedal On_R | 002-127 |. 060 .| R percussion noise when sustain/damper is released

The note range utilised for the GrandPiano samples is C1 to D#9. You will see that the "pedal on" event is mapped internally to the note E9 and the "pedal off" event is mapped internally to the note G9. It seems likely that the new "half pedal" facility will reduce the level of the piano resonance sound "Tailpiece" to approximately half the normal level of 127. This will be tested when a Korg DS-1H peadl arrives later this month.

There are one or two minor "improvements" that can be made:

a) Even when you try to play very quietly, the actual sample volume is played at a loud level (115).
b) The FX Key Off_L sound is not audible when you release the keys.
c) The piano sounds too "chorused" compared with a real Grand Piano.

To fix problems (a) and (b), you need to modify the OSC Level settings for some of the oscillators/ multisamples. All you do is go into the Sound -> Basic -> OSC Basic screen, and for the High OSC Multisample, touch the "Level" parameter and adjust it with the wheel to the new Level values below. You have to make sure you've selected the correct OSC number from the right scrolling list of radio buttons first.

Osc | MultiSample .| Velocity| Level |. Usage
1 .. | GrandPiano_L .| 104-127 |. 127 .| L speaker piano note, played ff
2 .. | GrandPiano_R .| 104-127 |. 127 .| R speaker piano note, played ff
3 .. | GrandPiano_L .| 089-103 |. 115 .| L speaker piano note, played f
4 .. | GrandPiano_R .| 089-103 |. 115 .| R speaker piano note, played f
5 .. | GrandPiano_L .| 045-088 |. 100 .| L speaker piano note, played mf
6 .. | GrandPiano_R .| 045-088 |. 100 .| R speaker piano note, played mf
7 .. | GrandPiano_L .| 002-044 |. 080 .| L speaker piano note, played pp
8 .. | GrandPiano_R .| 002-044 |. 080 .| R speaker piano note, played pp
9 .. | Tailpiece_L ......| 002-127 |. 127 .| L speaker piano resonance if sustain/damper is on
10 .| Tailpiece_R ......| 002-127 |. 127 .| R speaker piano resonance if sustain/damper is on
11 .| FX Key Off_L ...| 002-127 |. 018 .| Percussion noise when you release a note
12 .| FX Pedal On_L | 002-127 |. 060 .| L percussion noise when sustain/damper is depressed
13 .| FX Pedal On_R | 002-127 |. 060 .| R percussion noise when sustain/damper is depressed
14 .| FX Pedal On_L | 002-127 |. 060 .| L percussion noise when sustain/damper is released
15 .| FX Pedal On_R | 002-127 |. 060 .| R percussion noise when sustain/damper is released

Then to fix problem (c) go into Sound -> Effects -> FX Select, and note that FX2 is currently set to "26 - Stereo Chorus". Change the FX type to "110 - PianoBody/Damper". Then set the FX parameter settings as follows:

SoundBoard Depth - 50
Damper Depth - 30
Src - Damper #64
Tone - 90
Mid Shape - 20
Tune - 0
FX Amount - 86
Src - Damper #80

After making these changes, don't forget to write the modified sound - for example with with the revised name "Grand Piano RX 2" to one of your User bank sounds.

If anyone wants a copy of the finished .SET file, you can download it here. You have to become a member of www.createsongstyles.com before you can see the downloads. But it only takes 10 minutes to make your own and you will learn a lot by doing it for yourself! Have a go

http://www.createsongstyles.com/forum/i ... pic=4770.0.

Lee · Post by **Lee** » Wed Apr 16, 2008 8:18 pm

Rob,
What you are teaching us is truely SUPER.....it's catalyst for us to learn a lot!
I want to eventually get into making some of my own voices/modifying factory ones.... but not a lot of time right now (full time job).

I will try these mods tonight! It will be fun and educational... I hope I can hear the difference like you can.

Along these lines I do have a general question.

Some of the solo type voices have very different vibrato speeds on different notes... Not so good for sure when you play slow music and that music has a lot of held notes. Some voices are fine. This is dissapointing to me as I play that kind of music mostly.

Is the problem in the samples? If so I guess were stuck with it?
Or can I fix these issues myself with the programming facility in the PA2?
Oh, by the way, I played a M3 the other day and it showed the same issue with some voices.

Thanks,
Lee

Rob Sherratt · Post by **Rob Sherratt** » Wed Apr 16, 2008 8:34 pm

Hi Lee,

Give me an example of a factory sound you would like to improve, tell me what improvements you feel it needs, then we can work on it together - maybe posting updates here so everyone can learn about it?

dimitris · Post by **dimitris** » Wed Apr 16, 2008 8:38 pm

Hi Rob,

By the way maybe you have forgotten to send me the "GRANDPIANORX" sound of PA2X to experiment with my PA800.

Regards,
Dimitris

Rob Sherratt · Post by **Rob Sherratt** » Wed Apr 16, 2008 9:33 pm

dimitris wrote:By the way maybe you have forgotten to send me the "GRANDPIANORX" sound of PA2X to experiment with my PA800.

Just emailed it to you now. Sorry about the delay.

Lee · Post by **Lee** » Wed Apr 16, 2008 9:43 pm

OK, sounds great...

here it is: SAX group, Jazz Tenor 1

Example: Keys A5 ,B5
also C5, D5, E5 (other notes too, but this is a good example)

The vibtato is very different. Many other sounds are like this also.

I'm hopeful there is a solution.. also, the voices that use delayed vibrato controlled by aftertouch seem to be great!

Thanks,
Lee

galettouille · Post by **galettouille** » Wed Apr 16, 2008 9:53 pm

Hi Lee. The vibrato is included in the sample. It's not a "sound" parameter. Si it's not possible to change this.

Best
Alex

Lee · Post by **Lee** » Wed Apr 16, 2008 10:04 pm

Well... that's very dissappointing. The reason DSP is in there is to add vibrato etc, etc..

It should not be in the sample. In my opinion... cause how would you ever get all the samples to have the same vibrato? A slight difference is musically OK, and normal, but some of these notes are much, much different. The Korg M3 is the same way on some of it's voices, I played one yesterday. There are quite a few voices that exibit this issue.

It would be good for the kind folks at Korg to remidy this?

I guess we have to just not use those voices, at least for any music that's not fast all the time.

Lee

galettouille · Post by **galettouille** » Wed Apr 16, 2008 10:25 pm

In acoustic instruments the name "vibrato" is used to describe several simultaneous things. A sax "vibrato" is a combination of amplitude, harmonic, and pitch, variations. It's very difficult to imitate with synth parameters, that's the reason why more and more sounds are sampled with vibrato.
Best
Alex

dimitris · Post by **dimitris** » Thu Apr 17, 2008 9:44 am

Rob Sherratt wrote:
dimitris wrote:By the way maybe you have forgotten to send me the "GRANDPIANORX" sound of PA2X to experiment with my PA800.
Just emailed it to you now. Sorry about the delay.

Thanks, Rob! Ευχαριστώ!

Δημήτρης

shrike · Post by **shrike** » Thu Apr 17, 2008 10:38 pm

The vibtato is very different. Many other sounds are like this also.

Lee,

This issue is present on all Korg keyboards (probably on other brands too, but I don't wont to say that for sure because I own and play only Korgs). My guess is that even sound engineers don't create preset multisamples with one-key-one-sample technique, but they rather use algorithms to create upper and lower samples from one basic sample, much like one is able to do with Korg sampler in Triton, M3 etc.

So I guess that's why vibrato isn't equally fast on all keys in preset multisample (no matter which preset multisample, try any that simulates vibrato).

Lee, you can try it for your self and you'll hear it. Sample some sound with vibrato (even your voice, for example, simple AAAAAA sound with vibrato technique) in C5 pitch (or any other). Assign that sample in sampler to C5 key and let the sampler calculate three tones up and down (B4, A#4, A4 - C#5, D5, D#5). Now, listen the result. Only originally pitched sample will have the exact vibrato speed you produced during sample creation, but those other samples will have slightly different, slower or faster, vibrato.

Now, this is probably what sound engineers do with preset samples and multisamples. My guess is that if they would use one-key-one-sample technique, present ROM wouldn't support as many sounds as it does now (regarding which keyboard is at hand, ROM size vary between Triton, M3 or arrangers), so they would have to use bigger ROM or ROM wouldn't support as many multisamples as it does now. This gets even more complicated considering the fact that each sample has to have some reasonable period of duration to reach it's full character and then be looped after that point - otherwise, they wouldn't be useful.

Other fact to prove my statement: try listening preset multisamples in some initialized program slot, meaning without any parameters and effects set up. Listen very carefully and with good earphones. You'll hear there are groups of two, three and four samples with same character. For example, take some tenor sax preset multisample. Start with the C5 (this are examples and don't necessary relate to real thing) - C5, C#5 and B4 have one character of sounding. D5 sounds a bit different then those three, but close enough. D#5, E5, F5 and F#5 sound with same character as D5, but G5 sounds a bit different then those previous five. So you can conclude E5 is truly sampled and D5, D#5, F5 and F#5 are calculated from the original E5. In this example, G5 is probably derivated from some tone above itself.

Because of this, I can make no other conclusion - Korg isn't taking all the samples in one multisample, but some reduced number of them, according to possibilities to calculate them not to sound too different. This differences aren't easily audible, but for trained ear they are present.

The very same thing goes for user multisamples which consist of smaller number than 12 tones in one octave. Musicians are using multisamples structured as followed - one sampled sound and one or two calculated in each direction (one or two above the original sound and one or two bellow original sound). Or they sample only white keys and let the sampler calculate black keys, which is better solution, but not perfect.

It would be good for the kind folks at Korg to remidy this?

Maybe one day. I hope.

Lee · Post by **Lee** » Fri Apr 18, 2008 2:56 am

shrike,
Thanks for the information. I actually knew that they usually don't have the money or memory to use a different sample for each note. On some truly great sampled pianos I have they in fact did a triple strike... meaning they actually have 3 samples for each of 88 notes. You couldn't afford to do that on every sound.

But, here is the thing.. a large amount of the sounds in the PA2 and M3 sound terrific and the vibrato is fine.I assume they did not use samples that included vibrato.

Sampling the sound with vibrato is not a good idea because the speed will vary too much if you then use that sample (stretch) over too many notes. But it is easier and requires less programming for the sound engineers to just get the vibrato in the sample. At least I assume that is why they do it because the vibrato is not any better comming from a sample than DSP. If they are trying to get other nuances in the vibrato at the cost of a greatly varying speed, I would personally rather have a constant speed.

Actually if the vibrato changes proportionly it would be OK, but you go play along and all of a sudden a big change. Like c5,--f5 ok then g5 twice as fast, then a5 back to slower again etc....

When you play a song and you hold a note that and it's vibrato is a lot faster or slower than the speed of the note right next to it... sounds terrible. If you play only fast lines not much hoding notes then it's not any issue at all. It doesn't have to be perfect but if it's way off, it ruins the performance.
And of course it only matters with voices played for solo melody.

Why not make the vibrato come from DSP after the fact. A good example is... some of the sound 'groups' in the PA2 that I am talking about have voices that work superbly on all notes. So when you play those slow passages and hold notes they work great, usually with vibrato controlled with aftertouch. I enjoy them very much. There is a couple of Sax's that have touch vibrato that are really super to play.

I do think some of this exists in some other instruments, but it does not in my Kurzweil K2600 that I can remember, so it does not have to be.

There are lots of great voices to play...some of the ones with the sound I like, but the vibrato problem I just skip for now.

Lee

Rob Sherratt · Post by **Rob Sherratt** » Fri Apr 18, 2008 5:46 am

Hi Lee,

If it's not too much trouble, please can you send me a list of the sounds with "good" vibrato, and a list of those with "bad" vibrato? I would like to follow this up off-line and see if a solution might be possible.

thomasteele · Post by **thomasteele** » Sat Apr 19, 2008 8:37 am

Hi Rob
Could you please send me the piano sound , the src damper reads 64 on 800 model ,I cant get 80

Lee · Post by **Lee** » Sat Apr 19, 2008 5:47 pm

Rob,
Will do...Let's see what we can learn...
If you can learn something new each day it has been a good day!
Thanks,
Lee