Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards The DPBSD Project!

RE: Pianoteq, one can clip the velocity response at any desired threshold. To match it to my controller, I flat-line the response at 0 until a velocity of 8. (my Kawai MP9000 produces a minimum velocity of 7, so I want all velocities <= 7 to produce no sound, which incidentally, matches it's own response)

Note also that one can define a completely arbitrary velocity response in Pianoteq - any curve can be drawn.

Greg.

Yes, this is key.
I would expect more digital pianos to provide VPiano style functionality of making these kind of personal adjustments.

Kawai CA63 Re-Review

Due to all the recent interest in the Kawai CA63 I though I would take this opportunity to revisit the DPBSD MP3 kindly provided by kawaian to hone the existing review - in particular to quantify the loop lengths - plus capture some pictures of the analysis.

As I said before, this DP has nice things going for it. I can't see much in the way of stretching, and can't hear it at all as it only appears to exist on a few of the upper notes. Velocity layer blending is very well done, with only one step rather audible.

Attack sample lengths are about average, but the loop samples could benefit from some added length. I can hear C4 and C5 looping, but overall looping is fairly well done for such short loop samples - rather like the Yamaha CP1, which also has standard attack sample lengths and short loop lengths, but manages to sound pretty good (for something looped).

The notes decay a bit fast, though the quantization noise in the sample that keeps me from hearing everything going on near the noise floor.

If anyone can provide a DPBSD MP3 with a better levels, and particularly one that captures the key and pedal noises and sympathetic resonance, I'd be highly appreciative.

More pictures of this analysis, the MP3, and all reviews are at the share point.



Spectral phase view of the entire looping test, where the attack and loop sections of the lower note samples are clearly evident.


Spectral pan view of the note C2. Attack sample is 3.0 seconds, loop sample is 1.3 seconds, cursor located at the transition between attack and loop.


Spectral phase view of the note C4. Attack sample is 1.6 seconds, loop sample is 0.83 seconds, cursor located at the transition between attack and loop.


Spectral frequency view of the entire velocity layer test. The velocity switch three blocks in from the right is clearly visible and audible, but the rest of the range is nicely blended.


Spectral pan view of the mid section of the stretching test. Notes appear to be individually sampled. Low and high sections are similar, with the exception following.


Spectral pan view of the stretching test zoomed up. The group of four notes in the center and the group of two notes adjacent on left appear stretched to me, but it isn't audible. These are the notes A7 thru D8.


--------------
- Kawai CA63 -
--------------
FILE & SETUP:
- dp_bsd_v1.3_kawai_ca63.mp3
- Recorded by "kawaian".
PROS:
- Large dynamic range (~51dB, vel=1:127).
- No audible sample stretching.
- This is a very smoothly blended multi-velocity layer sample set.
- There are two semi-visible velocity switches @ vel=46,124.
- Responds to partial pedaling.
CONS:
- Looped, though not too badly done, would benefit from longer loop samples.
- Looping somewhat audible, particularly C4 & C5.
- Attack sample lengths are (C2:C9) 3.0,3.0,2.1,1.6,1.4,1.5,0.97,? seconds.
- Loop sample lengths are (C2:C9) 1.4,1.3,0.68,0.83,0.64,0.74,0.63,? seconds.
- Note decay somewhat short (~2/3 Pianoteq).
- One velocity layer switch is fairly audible (timbre change) @ vel=124.
- Stretch distances: 1(x72),2,4,1(x10) = 84 groups.
- No obvious pedal up/down or key up samples.
- No obvious sympathetic resonance.
OTHER:
- MP3 levels fair: peak @ -12dB, noise floor @ -80dB, significant noise @ floor.
- Date reviewed: 2010-02-16; updated 2010-03-17


[EDITS]
2010-12-15:
- Fixed note numbers in picture captions: C3=>C2, C5=>C4 (also fixed numbering in zip file).

I also updated the review for the Yamaha P-155 and posted a bunch of analysis pix at the share point.

Just btw, on my Kawai MP9000, I notice that on the Rhodes preset at least, the note evolves in discrete steps - the initial portion including the attack, a looped sustain (with an amplitude envelope), and then a final decay. I.e - the looped portion does not persist until the very and. The decay portion begins BEFORE the note is released - I am not referring to the release. I haven't determined whether this method us used for other presets yet.

Greg.

Greg,

I had a MP9000 a few years back. Great keyboard but I found the piano voice a bit lifeless. As I recall Kawai had nothing to do with the EP patches...they bought them from Wizoo or some company with a similar kind of name.

Cheers,

Steve

Hi James,

I had a 11-year-old Yamaha CVP96 and this did it the same way.
This is however only the case for the piano voices as it should be.

Currently I have a Kawai CP136 and it does something similar for some e-piano or harpsichord voices, but -oddly- does not do it for the piano voices.

I also have the GalaxyII piano library and if the option "silent key" is chosen, it does it the same way.
It also has a free form velocity curve, and if the option "Silent key" is chosen, any threshhold can be adjusted, using a flat curve start.
The same should be the case for Galaxy Vintage D.

Also the piano library "Pianissimo" supresses all velocities below 5.
So far I know, Pianoteq and Truepianos behave similar.

Best,

Peter

Steve,
RE: the MP9000, I think the acoustic sounds are actually not too bad (it's a very rich tone), although it is no match to the sounds I have on my computer. That's interesting about the electric pianos - I didn't know that. They're pretty good too, but let down by poor velocity mapping to the keyboard, IMHO. The Wurly completely falls to pieces in the bass registers - it seems to me that Kawai's harmonic imaging isn't (or wasn't, back then) good enough to reproduce the very rich timbre of the Wurly in the bass notes. It's pretty good elsewhere though.

Greg.

Yamaha P-155 Re-review

I see a lot of download activity for the P-155 DPBSD MP3, so I thought I'd take this opportunity to briefly re-review it and add a few pix. The text review was updated a few days ago too. Thanks again to ChrisA for the DPBSD MP3!

The attack and loop samples are rather short which makes the looping rather audible, and there is a fair amount of audible stretching going on over the entire range. I think I would have preferred they sacrifice a velocity layer in exchange for longer samples and less stretching. On the plus side the decay times are fairly long, and the velocity layers are nicely blended, with no visual or audible steps.

More pix, reviews, and MP3s at the share point.


Spectral phase view of the the entire looping test. Attack and loop samples of the lower notes clearly visible here.


Spectral pan view of the the note C4. The cursor is located at the transition between the 1.87 second attack sample and the 0.54 loop sample.


Spectral pan view of the the note C5. The cursor is located at the transition between the 1.7 second attack sample and the 0.59 loop sample.


Spectral pan view of the the note C6. The cursor is located at the transition between the 1.6 second attack sample and the 0.42 loop sample.


Spectral pan view of the the upper notes in the stretch test. Stretching is obvious, both visually and audibly, even in the higher registers.


Spectral frequency view of the the entire velocity layer test. All layers are smoothly blended, no steps apparent either visually or audibly.


----------------
- Yamaha P-155 -
----------------
FILE & SETUP:
- dp_bsd_v1.3_yamaha_p155.mp3
- Line out, line in to Mac, Garage Band.
- Recorded by "ChrisA".
PROS:
- A very smoothly blended 4 layer (reportedly) sample set with no visible or audible layer switching.
- Huge dynamic range (~62dB, vel=1:127).
- Sympathetic resonance.
- Responds to partial pedaling.
CONS:
- Obviously looped, could benefit from longer attack and loop samples.
- Attack sample lengths are: 2.4,1.9,1.87,1.7,1.6,1.2,0.8,? seconds.
- Loop sample lengths are: 0.7,0.78,0.54,0.59,0.42,?,?,? seconds.
- Visibly and audibly stretched over entire range.
- Stretch distances: 3,3,4(x3),2,4,3(x3),2,3(x3),2(x4),3,1,3,3,4,2,2,3,5,10 = 28 groups.
- No key-up or pedal samples or sound effects.
- Sympathetic resonance fairly fake sounding - echoy and reverby, particularly during parital pedaling.
OTHER:
- MP3 levels: peak @ -0.1dB, noise floor @ -69dB.
- Date reviewed: 2010-02-08; updated: 2010-03-17

FYI, I'll be away from the computer for a week or so, back to reviewing when I get back.

Job interviews in Japan, perhaps?

Cheers,
James
x

Either that or spending some time at a court house???
(Hope you can take a joke, Dewster!)

Or perhaps he's spending a week putting an embedded PC into his P-120. :P

Greg.

Ha! Nothing nearly as interesting as all that, just tending to some family business.

Someone at the marketing department has been here...

I don't remember if Dewster's analyzed any of the Roland models that has the supernatural sound. If yes, whether it's passed with flying colors like it should according to the YouTube infomercial above?

It passed with flying colours if I remember correctly.

You can download dewster's "dpbsd_reviews.txt" file, but below I've pasted his opinion of the 3 Roland DP's (v-piano, HP-307, and the RD-700GX with SN expansion card):

------------------
- Roland V-Piano -
------------------
FILE & SETUP:
- dp_bsd_v1.3_roland_v-piano_vintage1.mp3
- S/PDIF to CD recorder.
- Factory settings except gain +12dB & reverb off.
- Recorded by "EssBrace/Steve".
PROS:
- Beautiful long note decay.
- No looping (modeled).
- No stretching (modeled).
- Sympathetic resonance, though very subtle.
- Responds to partial pedaling.
- Good dynamic range (~36dB, vel=1:127).
- Nice "loom of strings" pedal down sample, pedal up sample more of a knock.
CONS:
- Key-up sample not audible.
OTHER:
- Quickly repeated notes have somewhat random velocities (modeled reality?).
- Date reviewed: 2010-02-09



-----------------
- Roland HP-307 -
-----------------
FILE & SETUP:
- dp_bsd_v1.4_Roland_HP-307.mp3
- roland_hp-307_key-off_mono.mp3
- roland_hp-307_resonance_0.mp3
- roland_hp-307_resonance_5.mp3
- roland_hp-307_resonance_10.mp3
- roland_hp-307_resonance_more.mp3
- roland_hp-307_resonance_more10.mp3
- Asus Xonar DX line-in, Adobe Audition.
- Recorded by "zaba19".
PROS:
- Beautiful long natural-sounding note decay (decay times on the order of Pianoteq).
- Large dynamic range (~47dB, vel=1:127).
- No audible looping.
- Visually, the notes look similar to conventional length attack samples seamlessly blended with something like long loop samples. Not clear what the process is.
- No visible or audible stretching, notes look random in the wave and phase views.
- No visible or audible velocity switching.
- Very smoothly blended timbre variation with velocity.
- Responds to partial pedaling, centerpoint of this effect is with pedal mostly down.
- Key up sounds like realistic string damping rather than a knock.
- Sympathetic resonance, sounds pretty good even when turned up.
- "Loom of strings" pedal down sample.
CONS:
- Can't detect key up, pedal up/down, or sympathetic resonance sounds in DPBSD MP3 file for some reason.
OTHER:
- Probably good enough to realistically record solo.
- MP3 levels: peak @ -13.5dB, noise floor @ -90dB.
- Rather large quantization noise in MP3.
- Date reviewed: 2010-02-17; review updated: 2010-03-01.


---------------------------------------------
- Roland RD-700GX with K-RD700GX1 expansion -
---------------------------------------------
FILE & SETUP:
- dp_bsd_v1.4_Roland_RD-700GX_K-RD700GX1_SuperNATURAL-Grand_Piano_no_sympres.mp3
- Recorded by "sandord".
PROS:
- Beautiful long natural-sounding note decay (decay times on the order of Pianoteq).
- Large dynamic range (~47dB, vel=1:127).
- No audible looping.
- Visually, the notes look similar to conventional length attack samples seamlessly blended with something like long loop samples. Not clear what the process is.
- No visible or audible stretching, notes look random in the wave and phase views.
- No visible or audible velocity switching.
- Very smoothly blended timbre variation with velocity.
- Responds to partial pedaling, centerpoint of this effect is with pedal mostly down.
- Pedal up sounds like realistic string damping rather than a knock.
- Sympathetic resonance (in-line effect?).
CONS:
- Can't detect key up or pedal down sounds in DPBSD MP3 file for some reason.
OTHER:
- Probably good enough to realistically record solo.
- MP3 levels: peak @ -3dB, noise floor @ -81dB.
- Some gated digital bleed-thru at the noise floor that comes and goes (PC?).
- Date reviewed: 2010-03-01.

Who's the babe?

Dewster:

Haven't read all the posts to see if you're accepting more mp3s, but this is another version of Pianoteq using the latest piano.

http://www.box.net/shared/slm031ovs5