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

Here is a dpbsd-v2.0 recording on a factory reset Roland HP-603 with SuperNatural Piano Modeling. I hope, I have set up the recording process correctly.

Compared to the sound experience directly in front of the HP603, the WAV recording sounds more raspy (on my Adam A5 monitor with a Tascam US-100 audio interface).

Dropbox: dpbsd-v2.0-Roland-HP-603...wav.7z

Hello lophiomys,

Your HP 603 recording is no longer available. Could you please reupload it?

Thank and regards!

Hello Semiactive,
Sorry, I have re-organised my files on Dropbox, not being aware, that it would destroy most of the links in the forums!

Re-uploaded the DPBSD v2.0 mid file of the Roland HP603:
HP603 recording of DPBSD-v2.0.mid

Thank you, Lophiomys.

If it is OK to put up the results of some freebie SFZ / Kontakt pianos

Piano in 162 from Ivy Audio
Steinway Model B grand piano, five velocity layers, two round robins, pedal on/off
http://www.mediafire.com/download/ti4d820tf6q8qxh/Piano_in_162_close_dpbsd.wav
http://www.mediafire.com/download/090s4h3vy0r01yk/Piano_in_162_ambient_dpbsd.wav

Recut Salamander Piano (tightened to improve response) created by Alexander Holmfeel, Yamaha C5 grand
Every third note sampled, 16 velocity layers
http://www.mediafire.com/download/sshhsfjus93acy4/Salamander_dpbsd.wav

Iowa Piano created by the University of Iowa Electronic Music Studios
Steinway & Sons model B
Three velocity layers, almost every note sampled.
http://www.mediafire.com/download/cfzae8cek7i2tfg/Iowa_dpbsd.wav

Summer Piano from the OLPC project by Dr. Mikhail Krishtal Director of Music Research and Production and his team at Zenph Studios.
Yamaha Disklavier Pro Piano
Every third note sampled, 10 to 15 velocity layers, pedal up/down
http://www.mediafire.com/download/w31km3dbqjpzb8h/Summer_dpbsd.wav

City Piano, Baldwin
Four Velocity layers, every note sampled
http://www.mediafire.com/download/9zc1kwgf76s0gbs/City_Joe_dpbsd.wav (dirty)
http://www.mediafire.com/download/3bul4kacz6yjkfn/City_Vospi_dpbsd.wav (clean)

Not sure if we can call this project "active" but I know how to analyze the MP3 files because dewster teach me long ago. But I am only interested on this models:

• Yamaha P115
• Korg SP 170S
• Korg B1
• Korg SP 280
• Korg Kross 88
• Roland FP20
• Roland FP30
• M-Audio Accent
• Alesis Coda
• Kurzweil KA90
• Kurzweil MPS10
• Kurzweil MPS20
• Kurzweil SP4-8
• Lowrey EZP3

If you have any of this models, please upload the MP3 or WAVE. I want this because in my project I compare digital pianos and show information that I can only know with this test MIDI file that dewster made.

Send me a personal message if you upload the test file of any of the models mention here.

Below link for DPBSD 2.0 test of Roland FP-30.
Includes manual resonance testing, noise level -56 dB

FP-30 test files

Thank you very much. I will work with those files as soon as possible and publish my results. I don't go as deep as dewster, but should be enough.

May I ask, what is the manual resonance test?

See the dpbsd_readme.txt:
Some DPs (esp. Kawai, Roland) lack pedal and/or key sympathetic resonance when played via MIDI. The following are two manual tests for these features.

Over five million views. Amazing. Dewster, we salute you, wherever you are. And please come back...

+1.

Salute him? For what?

The entire thread can be condensed into a simple statement:
Digital pianos have poor sound because:
1. The sounds are looped.
2. They lack resonance.
3. They lack sustain.
etc. etc. etc.
The same is true for them all. Perhaps that's why interest has waned?

A similar thread might say:
I found a blade of grass. It is green.
Here's another blade. It, too, is green.
Here's another. Green.
And another. Green.
Need this be continued?

Actually, there are some newer DP's which have un-looped sounds, reasonable sustain, and reasonable resonance (e.g. the Roland LX-17 and "SuperNatural modelled" cousins). They're not cheap (yet) -- outside Daniel's list of models.

Pianoteq 5 (IMHO) also does pretty well in those departments. It may sound "artificial", but that's a subjective judgement, not a quantitative one.

If Dewster wanted to come out of retirement (that is, spend more time _not_ earning a living<g>), there's some worthwhile testing he could do.

MacMacMac,

Dewster's work is very valuable!
It set things straight against overboarding advertisements, not by hearsay, but by hard facts.

If I would only have found the information in this thread, before I bought my digital, that would have been very helpful.
I do salute Dewster.

Kind Regards,
Lo PHi

It's not so much that I disagree with you Mac, but it was always a matter of degree. Each piano has/had its own characteristics. The note stretching for example...Kawai used 88 note sampling (as did Yamaha some years ago).

Likewise, some looping was done better than others. Nord uses loops but each loop is long enough to wobble in its own right before being repeated and that disguises the presence of looping to some extent - far better than the very short, static cycles employed by Yamaha and Kawai for instance, that both sound completely unnatural when decaying (IMO).

I think Dewster presented in graphical form what the listener/player hears and he provided a really interesting basis for comparison. And given that very few of us have played all the pianos featured in the DPBSD I do think the thread was, and remains, a very useful resource. To my mind it doesn't replace subjective testing but it does help to understand what is going on beneath the marketing blurb pedalled by the makers.

And as Charles says, there are pianos that have broken out of the tiny incremental change/improvement regime...Dexibell claim much longer attack samples and the modelled pianos (Roland, Physis) have no loops at all and no velocity layers.

Interest has waned simply because the library of results remains more or less at it was when Dewster left. If he was still testing pianos the thread would have remained current and, I suspect, well subscribed to.

Just my thoughts on the matter...

If I remember correctly, dewster was a former telecommunications engineer who had left his company and was working on his own research. Could it be that he found another job, and therefore no longer has time to participate on the forum?

For what it's worth, I personally found the DPBSD project to be very interesting.

Cheers,
James
x

My concern with DPBSD is that it's not very useful.

The tests and charts and graphs mean little by themselves. But given an explanation, they all point to the same thing: mediocre sound performance.

That conclusion could more easily be reached by simply putting fingers on the keys.

Yes, there are differences in the sounds of the various pianos. But the ears tell the story much better than the data.

You miss the point. DPBSD is only concerned with sound generator issues. By putting the fingers on the keys you mash another complication into the picture: The keyboard action. But at least for this project, keyboard action issues are not the focus.

DPBSD works via pure MIDI input and makes sound generators comparable, even if their source pianos offer vastly different keyboard actions - or no keyboard action at all, like expanders or virtual instruments.

I don't agree. Rather ... you've diverted around the point.

Which is ... My ears (with fingers engaged) tell me all the same things (and more) that the DPBSD tells me without fingers.

Your ears might be really good, and give you satisfactory perception. That might not be the case for many others.

The DPBSD project gives you quantitative and comparable measuring results, which are hardly refutable by marketing statements or personal preferences.