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

Ah, I see.

Thank you for the explanation.

Cheers,
James
x

I might believe it is a ground loop or "digital hash" on the ground. But it's not the signal in the USB cable. The signal is not RS232 levels. USB uses two twisted pairs and each pair uses differential signalling and very little current. The differential drivers and receivers on each end do something like an optical isolation would do on an older system.

If I'm right then a good size ferrite toroid on each end of the MIDI cable might help a lot. Get one where you can loop the cable through multiple times.

Yamaha labeling it "to host" does sound confusingly USBish. It's a weird interface Yamaha dreamed up at some point I guess. I made a cable myself from an old broken hand-held scanner (for the barrel connector molded to a length of cable) and a 9 pin DSUB connector.

Plugs into my PC serial port. 100% RS232, which employs single-ended electrical signaling, which means there are ground currents, which means noise on the audio ground unfortunately.

Hi, here is the test file of Casio PX-330. I've added also Purgatory Creek demo. I hope these links will work.

http://www.mediafire.com/file/rnm3wmmn13r/dp_bsd_v1.3_casio_px330.mp3

http://www.mediafire.com/file/dqyz5mnzhtu/Casio PX-330.mp3

Pardon my english but are You absolutely sure about that? 'Cause I did make a new Setup out of Concert Grand with Reverb OFF and Damper Resonance, String Resonance and KeyOff maxed to 10 and recorded MIDI file with it. When I playback the file with the same Setup all these effects can be heard or I might not fully understood what are You talking about on this matter.

R0B, may I ask which firmware version you are using on your MP5?

I'll have to confirm this point with the R&D chaps on Monday, however there have been a handful of updates to the MP5, so it's quite possible that the string/damper resonance over MIDI implementation has been added. This might also explain why batak would appear to be hearing the resonances.

I don't wish to get your hopes up, however please do check http://kawai.de/downloads_en.htm and consider downloading the latest software update (v1.15).

Cheers,
James
x

@batak,

here you can find several issues with the CA-51, CA-71, CA-91, among them also the MIDI Resonance Problem (unfortunately it's in German), and the MP-5 was no exception, I've tested that some time ago at my local dealer. I need to mention however that some of the issues are already fixed by newer firmware releases, so I would recommend to always install the newest firmware.

http://www.razyboard.com/system/morethread-ca91-kawai-122539-4944834-0.html

Hi James,

Many thanks for pointing me to the firmware update :-)

It has been some time since I updated, and found I was using Ver 1.14.

I have now updated to the latest, Ver 1.15, and will try it out tomorrow.

Thanks again :-)

Kind regards,

Rob

Thanks very much for those pesk! The levels are very nice. I noticed the E6 note did not play for some reason?

The PX-330 has somewhat shorter samples than Yamaha or KAWAI, and the decays of the lower notes seem faster as well. The looping is not quite as nice either, with a shorter crossfade. It does have the best velocity layer blend I've heard so far, nicely spread out and with no visible or audible timbre steps.

-----------------
- Privia PX-330 -
-----------------
FILE & SETUP:
- dp_bsd_v1.3_casio_px330.mp3
PROS:
- Note decay is fairly long.
- Large dynamic range (~49dB, vel=1:127).
- This is a VERY smoothly blended multi-velocity layer sample set (Casio reports 4 layer).
- Responds to partial pedaling.
- Something going on with sympathetic resonance, but it's so subtle I can't hear it.
CONS:
- Initial low note decay is rather quick, with a leveling out later in the decay.
- Obviously looped, not quite as well done as Yamaha or Kawai.
- Samples are 2 to 0.8 seconds (lo to hi) with crossfade.
- Obviously stretched, group transistions farily audible.
- Stretch distances: 3,2,3(x16),4,2,3(x5),4,5,5 = 28 groups.
- No obvious pedal up/down or key up samples.
OTHER:
- Note E6 didn't play for some reason.
- Volume in MP3 file very good.
- Date reviewed: 2010-02-12

Not trying to blow my own horn here, but for me anyway this project has been a real eye-opener. It has revealed the fact that KAWAI, Yamaha, and Casio can all do velocity layer blending well, they all incorporate roughly the same amount of note stretching, and their actual sample lengths are in the same ballpark.

It's also helped me nail down what were previously just vague impressions from demoing the various DPs in noisy stores. So far, my personal perception of DP sound quality appears to be highly correlated to actual sample length, how well looping is implemented, and rate of note decay.

In short, if a DP can't play individual notes fairly convincingly, then it doesn't sound very realistic to me when playing real pieces.

I know everyone is now interested in the new CP-series and SuperNatural pianos, but HERE are two samples from Roland RD-700SX: 'Superior Grand' and 'X Ultimate'. Both are labeled as 88-sampled. Reverb is off, and default settings are used as would be it if you play the piano out of the box. The default setting for damper resonance is barely audible but if you like, I can reupload new recordings with the effect at max level. The piano also has string resonance feature which is off by default and I have left it off as well for the recording. Recording has been made in Sony Sound Forge 8, with Cubase SX as MIDI-player (at tempo 120), the main volume slider on the piano was at 80%, the wavs were then peak normalized to -1dB. The sound card is Creative Audigy. The cables were not quality ones and I have used USB-cable, not MIDI.

Dewster, please tell me when you have downloaded the files from the link above, in order to free hosting space.

I will wait for your analysis but to my ears both pianos have 4-layers, no stretching, obvious looping, obvious layer switching, almost nonexistent sympathetic resonance and they both respond to half-pedaling. The X-Ultimate is very short-decaying sound and I have never used it because the transition from ppp to fff has nasty jumpy quality I hate.

Thanks for review dewster. I have no idea why E6 was not played but during recording test I've recognized that sometime some note was dropped out. Maybe it's issue of my MIDI setup...

CyberGene, I can't seem to get to that web page (it times out).

Works for me!

Works for me now too... got them!

I think you should blow your own horn, digital or otherwise. This is a remarkably useful exercise, Dewster, and could be used for many years to come as a reference point for serious purchasers.

It may be a little early to try to correlate the objective analysis with the more subjective sound quality issues, but at the very least, as it becomes established, it will hold manufacturers to account over their marketing claims. In time, I think it will become clearer which elements, probably more through their interaction than as individual components, contribute to a "pleasing" experience for both listener and player. The big challenge will come when modeling becomes the rule rather than the exception, as the analysis will probably need to recognize much more subtle variations than velocity stepping or note stretching.

CyberGene, I really appreciate the files!

I agree with your evaluation except for the layer count. I believe the first audible layer transition in both is actually just a filter switch.

The decay rate on the X is way too fast, and the velocity transitions are quite abrupt and placed too high IMO. The sample lengths for the X are also rather short.

----------------------------------
- Roland RD-700SX Superior Grand -
----------------------------------
FILE & SETUP:
- dp_bsd_v1.3_roland_rd-700sx_superior_grand.mp3
- Cubase SX as MIDI-player, Sony Sound Forge 8, Creative Audigy.
- Factory settings & reverb off.
PROS:
- Very little evidence of stretching except (F#5,G5) - top octave or so is indeterminate.
- OK dynamic range (~31dB, vel=1:127).
- 3 layers (from spectral pan view) with some kind of filter switch @ v=38.
- Layer transitions at v=68,108.
- Responds to partial pedaling.
- Something going on with sympathetic resonance, but it's so subtle I can't hear it.
CONS:
- Note decay rate is rather fast (1/2 Pianoteq or less).
- Looped, mids and highs are not very well done (very "loopy").
- Samples are 3 to 0.3 seconds (lo to hi) with crossfade.
- Rather abrupt layer transitions.
- No pedal up/down or key up samples.
OTHER:
- Noise floor in file is rather high (-66dB) but otherwise fine.
- Date reviewed: 2010-02-13


------------------------------
- Roland RD-700SX X Ultimate -
------------------------------
FILE & SETUP:
- dp_bsd_v1.3_roland_rd-700sx_x_ultimate.mp3
- Cubase SX as MIDI-player, Sony Sound Forge 8, Creative Audigy.
- Factory settings & reverb off.
PROS:
- Very little evidence of stretching - top octave or so is indeterminate.
- 3 layers (from spectral pan view) with some kind of filter switch @ v=58.
- Layer transitions at v=98,118.
- OK dynamic range (~31dB, vel=1:127).
- Responds to partial pedaling.
CONS:
- Note decay rate is pretty fast (1/3 Pianoteq or less).
- Looped, most obvious on the very highest notes.
- Samples are 2 to 0.3 seconds (lo to hi) with crossfade.
- Rather abrupt layer transitions.
- No obvious sympathetic resonance.
- No pedal up/down or key up samples.
OTHER:
- Noise floor in file is rather high (-66dB) but otherwise fine.
- Date reviewed: 2010-02-13

Dewster, thanks for the review! By the way, I think once I had time to play with the fixed velocity function for each key in order to analyze how many layers are used and I remember I had concluded there are 2, 3 or 4 layers per key depending on the pitch (basses used more layers that highs), but I may be wrong and also there may be a filter as you say.

Sorry for the noise floor. I didn't have a cable, that's why I myself fabricated one from three different cables and couplings

To be honest, I haven't used the internal sounds of RD-700SX ever since I got Ivory Italian Grand.

Very good point. It's entirely possible they use different layers for different ranges, and it's something I don't test for. Not sure I really want to though, the layer test already drives me a little crazy.


No problem, and thanks very much for going to the trouble of recording them!


Not trying to harsh on your RD-700SX, but I can't say I blame you.

Lawrence just sent me an DPBSD MP3 of the CP1 he got just today - thanks Lawrence! He gave me the go-ahead to post it for him.

They've done a good job of blending the velocity layers, but I think we anticipated that. The low notes have nice decay, but the mid and high notes decay rather quickly. The sympathetic resonance is rather typical for Yamaha, on par with the P-155 IMO.

The actual sample lengths are what surprised me. They hide it pretty well by doing a good job on the looping, but I was expecting something longer than what I've seen in just about every other mid-level DP so far.

--------------
- Yamaha CP1 -
--------------
FILE & SETUP:
- bsd_v1.3_yamaha_cp1.mp3
- Default patch 1 CF Grand (CF 3Band) Piano + PreAmp + MasterEQ blocks enabled.
- Cubase
PROS:
- Lowest notes have nice long decay.
- Good dynamic range (~38.5dB, vel=1:127).
- I believe this is a very smoothly blended 4 layer sample set with no audible layer switching.
- Visible layer switch @ vel=54,96,120 (spectral pan & phase displays).
- Sympathetic resonance.
- Responds to partial pedaling.
CONS:
- Mids and highs have somewhat short decay.
- Looped, though well done.
- Samples lengths are (C2:C9) 3,2.2,2,1.8,1.7,1.3,1,0.8 seconds.
- Audibly stretched at the low and high ends, no stretching of mids.
- Stretch distances: 3,3,4,4,4,2,4,3,1(x26),2,3,2,1,1,1,3,2,2,3,4,2,3,3,3 = 49 groups.
- No obvious pedal up/down or key up samples.
- Sympathetic resonance is echoy and reverby.
OTHER:
- Volume and noise floor in MP3 file excellent.
- Date reviewed: 2010-02-13