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

Big thanks to dewster for his indepth reviews, time and patience heading this DPBSD Project!

Thank you mkhor! Without kind souls offering up their DPs for review this project would have ground to a halt long ago.

Also: I just made a couple of edits to the CVP-505 text review to clarify something. There definitely are key-up string samples that sound like the string being damped, and you can adjust this from 0-10. But there are no mechanical sounds that I could detect, such as key-up "knock" or pedal up/down "loom of strings".

Nord Piano Beg-A-Thon

The Nord Piano is really interesting. It's lightweight, compact, has relatively large (for a hardware DP) sample sets that you can pick and choose from, seems to have realistic sympathetic resonance and mechanical noises, etc.

The downsides are possibly the Fatar action (who knows, perhaps Nord is somehow specifying the keys and watching over the quality control?) and the crazy fire engine red paint job which many seem to dislike but nevertheless is a Nord standard.

Go see it here, and be sure to listen to the sound examples!

http://www.nordkeyboards.com/main.asp?tm=Products&clpm=Nord_Piano

I captured the sound samples while they were playing and took a look at them in Adobe Audition. I've got a couple of pix to show you here.



The beginning of the "string resonance off" sample, note E3 in isolation, spectral phase view.


Same as above, spectral pan view.

The note sounds really nice, actually all the samples there sound really nice. The problem for me and my analysis is that it doesn't last long enough to see any looping, or whatever Nord is doing with note decay. The note above lasts only 3 seconds or so, and with the larger than usual sample memory in the Nord Piano, I'm pretty sure that isn't past the attack sample.

Could someone who has access to a Nord Piano please perform the DPBSD test on it? Please? Preferably on the "Grand Lady D" large 78.6 MB patch? I promise to run the review by you first, and won't publish it here in the forum until you are 100% satisfied that it is complete and accurate.

Please?

Second Begging Post In A Row!

I'd really, really (really) appreciate anyone contributing DPBSD MP3s of the following:

- Nord Piano - Grand Lady D (the "large" version: 78.6 MB)
- Yamaha CP50 - CFIII grand (default piano)
- Yamaha CP1 or CP5 - S6 grand
- Casio Privia PX-130 - grand piano modern (default piano)
- Casio Privia PX-3 - grand piano 1 (default piano)
- Roland RD-700GXF - default SuperNATURAL piano with all resonance effects enabled

Any other suggestions? Any other DP or particular voice in a DP that interests you?

If anyone wants to redo any of the older DPBSD MP3 files with the very latest MIDI file (v1.8) that would be very cool. I would definitely update the reviews based on the newer files. I would particularly appreciate a redo on anything currently v1.5 and below.

Please know too that if you provide a DPBSD MP3, I will coordinate with you privately, showing you the review text, my comments, and any analysis pictures before the review is posted publicly. I want you to be 100% happy with how your DP gets reviewed.

Great project, keep on good job. Thank dewster who gives us a chance to see detailed reviews on most DPs.

It's getting kind of lonely on this thread. Anyone feel like doing a DPBSD MP3 on some random DP just for kicks? I'm getting rusty...

I have a Kawai ES6 and a Macbook, don't know if I can help !?

Great work dewster ! I enjoy reading the reviews.

Ok, so I can't provide you with anything on the wish list, but a while ago I
ran the dpbsd 1.8 MIDI file through my Roland Juno-Di keyboard, using the
"88StageGrand" sound, which is the default sound when it starts up.

The WAV-file was recorded with Audacity through the sound input on my iMac.

The 131 Mb WAV-file is available here, if it has any interest for you:
https://docs.google.com/uc?id=0B28m...ODEtMjU2ZDdlNmRkMmU4&export=download

Thanks jens4711!

Jose Hidalgo, I sent you a PM.

Roland JUNO-Di Review



PW member jens4711 recently provided us with a DPBSD file for the Roland JUNO-Di - thanks jens4711! With high portability, D-Beam, vocoder, and filter knobs, this looks like a very fun little keyboard!

I finished the analysis yesterday and as usual took some pix during the process. I also converted the WAV file to MP3. Both are located at the share point here:

pix: http://www.mediafire.com/?ff1io2dn4776igl
mp3: http://www.mediafire.com/?omfe3osenxv1pa4

I was fairly surprised that I couldn't easily see any stretching. Not to say it isn't stretched, there is some visual evidence of it, just not as much as I'm used to. It sounds stretched but without clear visual cues I'm reluctant to state how much.

I was pleasantly surprised to see that it supports partial pedaling and brief partial damping. I could easily see and hear the three velocity layers, though the switching is unblended and so rather abrupt. Looping is pretty audible, this instrument would benefit greatly from longer attack and loop samples. The decay time is quite short, perhaps to mask the obvious looping. There is no sympathetic resonance of any type that I could detect, and no extra noises like key-up clunk or pedal up/down string sounds.

Dynamic range is fairly narrow at 30dB, and as a result of this and notes played at velocity = 1 are pretty loud.

I noticed strange stereo pan differences between the attack and decay samples. Many notes start out somewhat panned left or right, and then switch to a more panned decay, which sounds rather odd in my headphones.

All-in-all not too shabby for a piano voice sharing a 64 MB ROM with many other instruments, and in what is marketed as a highly portable synthesizer rather than a digital piano. Thanks again jens4711!



Waveform view of the looping test with vertical zoom applied. Decay times are pretty short.


Spectral frequency view of the layer test (compressed 20:1). The three velocity layers are clearly seen here. No real timber variation within the lower two layers, and the steps between layers are quite audible.


Spectral pan view of the looping test, note C2. Attack sample and loop transition are clearly seen, cursor @ at the transition point. Both attack and loop samples are fairly short, looping is pretty audible.


Spectral phase view of the stretch test, mid notes. I can hear something that sounds like stretching going on, and I can see consecutive notes here that look similar, but not as similar as I'm used to seeing in a stretched instrument which makes it difficult to quantify.


------------------
- Roland JUNO-Di -
------------------
FILE & SETUP:
- dpbsd_v1.8_roland_juno-di.mp3
- "88StageGrand" patch (default).
- Recorded with Audacity on iMac, WAV (float) => MP3 conversion via Audition.
- Recorded by "jens4711".
PROS:
- Passes the pedal down silent replay test but the vel=1 note is audible.
- Passes the brief pedal partial damping test.
- Passes the partial pedaling test.
- Not much visual evidence of stretching (though it kind of sounds stretched).
- This is an unblended three velocity layer sample set with some filtering of the highest range.
CONS:
- Fails the pedal down sympathetic resonance test - no detectable sympathetic resonance.
- Fails the key down sympathetic resonance test.
- Notes played at vel=1 are pretty loud.
- Note decay has more pronounced stereo pan than note attack.
- Very short note decay time (~1/3 Pianoteq).
- Visibly and audibly looped.
- Attack sample lengths are (C1:C8): 1.6,2.0,1.6,1.3,1.4,0.8,0.6,? seconds.
- Loop sample lengths are (C1:C8): 1.2,0.5,0.5,0.6,0.4,0.3,0.15,? seconds.
- Velocity layer switches are visible and rather harshly audible @ vel=52,92.
OTHER:
- Dynamic range 30dB (vel=1:127).
- MP3 levels: peak @ -1.1dB, noise floor @ -74dB.
- Date reviewed: 2010-09-08.

Kawai MP5 Re-review



PW member kurtie just sent me an updated DPBSD file for the Kawai MP5 - thanks kurtie! The last version I reviewed was DPBSD v1.3, which lacked many of the tests in the current v1.8 file, so it's nice to be able to revisit the MP5 with a more complete review of its capabilities.

Pictures of the analysis and the MP3 can be found here:
pix: http://www.mediafire.com/?1929iiwnbe5c06k
mp3: http://www.mediafire.com/?or18kz1y7rd18ci

The looping here is pretty well done, particularly for a DP of this vintage. I don't like looping at all, but the implementation is less obnoxious than most. This allows the DP to have realistically long note decay, which is also very welcome.

The velocity layers are very nicely blended, I couldn't hear any timbre steps or static ranges, and timbre changes very smoothly with increased velocity.

String damping is also well done with key and pedal up. Partial pedaling and brief partial damping are present and behave correctly, though it fails the pedal down silent replay test at the end where the pedal is lifted while the key is still depressed - the note damps at this point and it shouldn't.

Pedal down sympathetic resonance is implemented, but the effect is subtle - it's clearly visible, but near the edge of my being able to hear it. Key down sympathetic resonance is also implemented, with harmonics of notes both lower and higher than the stimulus ringing out. Very nice.

The Achilles heel of this DP is in the stretching department. It is highly stretched - only 22 groups cover 88 notes with an oddly uniform 4 grouping all the way up the keyboard. The transitions between groups are fairly audible, particularly in the low and mid ranges, and I can hear some of the higher group transitions as well which is rather unusual.

I'm glad to see Kawai eliminating stretching in their latest offerings, it is the main sound issue holding back this DP.


Waveform view of the looping test with vertical zoom applied. Decay times are nice and long.


Spectral frequency view of the layer test. Nice, smooth timbre variation with increasing velocity.


Spectral phase view of the looping test, note C3. Attack and loop samples are clearly seen, cursor @ at the transition point.


Spectral pan view of the pedal down sympathetic resonance test, decay tails only (stimulus trimmed), pedal down at left of cursor, pedal up at right. The effect is quite visible, but audibly subtle.


Spectral phase view of the stretch test, mid notes. Groups of 4 all the way, baby. All group transitions are audible over the low and mid ranges, some over the high.


-------------
- Kawai MP5 -
-------------
FILE & SETUP:
- dpbsd_v1.8_kawai_mp5.mp3
- Recorded by "kurtie".
PROS:
- Passes the pedal down sympathetic resonance test - the effect is visible and somewhat audible.
- Passes the key down sympathetic resonance test - notes above and below the stimulus resonate.
- Nice long note decay on the order of Pianoteq.
- Passes the brief pedal partial damping test.
- Passes the partial pedaling test.
- Looping is pretty well done, particularly for an older DP.
- This is a very smoothly blended multi-velocity layer sample set (unknown layer count).
- Key-up & pedal up string damping sound is subtle yet realistic & pleasant.
CONS:
- Fails the pedal down silent replay test @ pedal up (string mutes).
- Visibly and somewhat audibly looped.
- Attack sample lengths are (C1:C8): 3.0,1.8,1.6,1.5,1.6,1.0,?,? seconds.
- Loop sample lengths are (C1:C8): 1.6,1.2,1.1,1.3,1.0,0.8,?,? seconds.
- Obviously stretched, group transitions are quite visible, lows and mids fairly audible.
- Stretch distances: 4(x22) = 22 groups.
- No obvious pedal down "loom of strings" or key up "knock" samples.
OTHER:
- Notes played at vel=1 are audible.
- Dynamic range 41dB (vel=1:127).
- MP3 levels: peak @ -4.8dB, noise floor @ -65dB.
- Date reviewed: 2010-09-11

Nord Piano Double Voice Review



Fellow PW member EssBrace recently ran his Nord Piano throught the DPBSD gauntlet, so we now have two MP3 samples of it - huge thanks to Steve! Ever since I learned of the downloadable voice libraries for the Nord products I've been anxious to review them. Going by the file sizes on the Nord web page, the sample sets are small compared to PC sampler offerings, but rather large by hardware DP standards, and are available in small, medium, and large versions, presumably to better manage Flash space in the the instrument. Steve provided DPBSD MP3s of the "Grand Lady D" which is a Steinway Model D, and of the "Studio Grand 2" which is a Yamah C7 - both from the largest sample sets of these instruments.

First off, the sympathetic resonance seems to be nicely implemented, I can hear a dispersive effect with the pedal down, and with the key down notes both below and above the stimulus ring out in sympathy. It seems one or more new DSP algorithms are sweeping through the industry lately, which is most welcome.

I can hear note release string damping samples, both at key and pedal up events, which sound realistic. Both pass the brief pedal partial damping test, a brief pedal damp causes somewhat more damping than with other DPs I've tested. Both pass the pedal down silent replay test, though the Steinway does better than the Yamaha here - I can hear the velocity = 1 note play in both, and some partial damping at pedal up in the Yamaha. There doesn't seem to be any support for half pedaling with either sample set, though there are pedal "clunk" and "loom of strings" sound effects. Steve reports that this DP senses damper pedal velocity, and pedal movement in the partial pedaling test is apparently too abrupt, so they sound rather loud and obnoxious in the test but probably work fine in actual use. Steve says the pedal behavior is quite advanced.

With a larger sample space to work with, one would expect looping issues to be minimized. And indeed this is the case, in both instruments the attack and loop samples are longer than most looped DPs, the attack is long enough to sound realistic, the blend to the loop is pushed out a bit, and the loops are long enough to "wobble". Still, the attack and loop sample sections are easily seen, and I can hear the looping in the low and mid notes. I also feel the loop "wobble" is a bit overdone, particularly in the Yamaha - it almost sounds like a slight phase effect. With the somewhat larger sample space they are able to provide a fairly long note decay time without overly revealing the looping.

With ~30 stretch groups, both instruments are pretty stretched, audibly so in the low and mid ranges, with significant timber variation between many of the lower stretch groups. In terms of velocity layers, if the Steinway was a PC ROMpler I would say it has 7 layers, as there are six visual and audible timbre steps consistent with an unblended sample set. But being such a small sample set I'm wondering if some of these are just filters kicking in/out? The Yamaha is smoothly blended with no cues as to the number of layers. In both sample sets the dynamic range is a bit shallow at ~33 dB, which causes notes at vel=1 to sound somewhat loud.

MP3s and all analysis pix here:
mp3: http://www.mediafire.com/?dnxq2695n73vv
pix: http://www.mediafire.com/?j7otn5u3d3k81

Some analysis pics and the text reviews follow. Thanks again to Steve for providing this intimate look at his new DP!


Spectral pan view of the pedal down sympatheic resonance test, stimulus removed, Yamaha is shown, Steinway is similar. On the left the pedal is down, on the right it is up. The effect sounds good to my ears.


Waveform view of the looping test with vertical zoom applied, Steinway is shown, Yamaha is similar. Decay times are nice and long.



Spectral pan view of the looping test, note C2, Steinway above, Yamaha below. Attack and loop samples are clearly seen, cursor is located at the transition point, looping is audible in the low and mid notes, the Yamaha loop "wobble" sounds more exaggerated to me.


Spectral pan view of the stretch test, mid notes, Yamaha is shown, Steinway is similar. A fair amount of stretching going on, audible in the lows and mids, with some groups sounding more muffled than their neighbors.



Spectral phase view of layer test, compressed 20:1, Steinway above, Yamaha below. Visual and audible evidence in the Steinway of 4 to 7 layers; the Yamaha is smoothly blended.


-------------------------------------------
- Nord Piano Grand Lady D (large 78.6 MB) -
-------------------------------------------
FILE & SETUP:
- dpbsd_v1.8_nord_grand_lady_d_steinway_model_d_large.mp3
- MacBook Pro (MIDI) => NP => stand alone CD writer => iTunes.
- Recorded by "EssBrace/Steve".
PROS:
- Passes the pedal down sympathetic resonance test - the effect is visible and audible.
- Pedal down sympathetic resonance is pleasant sounding.
- Passes the key down sympathetic resonance test - notes above and below the stimulus resonate.
- Passes the pedal down silent replay test, though with one issue (see CONS).
- Passes the brief pedal partial damping test.
- Pedal "clunk" and "loom of strings" sounds.
- Note release sounds @ key and pedal up.
- Fairly long note decay time (~3/4 Pianoteq).
- Attack and loop samples are longer than most looped DPs, loops are long enough to "wobble".
- This is an unblended 4 to 7 velocity layer sample set.
CONS:
- Pedal down silent replay test issue: vel=1 note play is audible.
- Fails the partial pedaling test.
- Partial pedaling causes loud pedal sound effects.
- Visibly and audibly looped, loop "wobble" is a bit overdone.
- Attack sample lengths are (C1:C8): 3.0,3.5,3.2,2.3,2.1,1.0,?,? seconds.
- Loop sample lengths are (C1:C8): 2.2,1.9,2.2,2.1,2.2,1.5,?,? seconds.
- Visibly and audibly stretched.
- All low & mid stretch group transitions fairly audible.
- Large timbre variations between many of the lower stretch groups, some sound muffled.
- Stretch distances: 2,3,2,3(x3),2,3,2,4,2,3(x3),4,3,2(x3),3(x3),4,4,2,4,3(x4),2 = 31 groups.
- Velocity timbre switches are visible and audible @ vel=48,66,86,98,108,116.
OTHER:
- Notes played @ vel=1 produce sound.
- Dynamic range 33.5dB (vel=1:127).
- MP3 levels: peak @ -1.1dB, noise floor @ -84dB.
- Date reviewed: 2010-09-14.


---------------------------------------------
- Nord Piano Studio Grand 2 (large 68.6 MB) -
---------------------------------------------
FILE & SETUP:
- dpbsd_v1.8_nord_studio_grand_2_yamah_c7_large.mp3
- MacBook Pro (MIDI) => NP => stand alone CD writer => iTunes.
- Recorded by "EssBrace/Steve".
PROS:
- Passes the pedal down sympathetic resonance test - the effect is visible and audible.
- Pedal down sympathetic resonance is pleasant sounding.
- Passes the key down sympathetic resonance test - notes above and below the stimulus resonate.
- Passes the pedal down silent replay test, though with some issues (see CONS).
- Passes the brief pedal partial damping test.
- Pedal "clunk" and "loom of strings" sounds.
- Note release sounds @ key and pedal up.
- Fairly long note decay time (~3/4 Pianoteq).
- Attack and loop samples are longer than most looped DPs, loops are long enough to "wobble".
- This is a blended sample set with no visible or audible layer switches.
CONS:
- Pedal down silent replay test issues: vel=1 note play is audible, partial damping at pedal up.
- Fails the partial pedaling test.
- Partial pedaling causes loud pedal sound effects.
- Visibly and audibly looped, loop "wobble" is a bit overdone.
- Attack sample lengths are (C1:C8): 3.5,3.1,3.2,2.1,1.9,0.9,?,? seconds.
- Loop sample lengths are (C1:C8): 1.9,1.8,1.8,1.5,1.4,1.1,?,? seconds.
- Visibly and audibly stretched.
- All low & mid stretch group transitions fairly audible.
- Large timbre variations between many of the lower stretch groups, some sound muffled.
- Stretch distances: 5,3,4,3,2,5,3,4,2,3,3,4,4,2,4,3,2,5,3,3,2,2,4,2,4,4,3 = 27 groups.
- Very little change in timbre with velocity over the lower ~1/3 velocity range.
OTHER:
- Notes played @ vel=1 produce sound.
- Dynamic range 32.3dB (vel=1:127).
- MP3 levels: peak @ -2.5dB, noise floor @ -84dB.
- Date reviewed: 2010-09-14.

Dewster,

Many thanks indeed for taking the time to analyse the performance of the Nord in the DPBSD test. On a purely subjective note I can say that the Nord is an unusually engaging and enjoyable thing to play. It has a definable character which can be very pleasing. Looking forward to the release of the Bosendorfer sample later this year!

Best wishes to all,

Steve

Thanks for the detailed review dewster, and of course congrats to Steve on his new Nord.

Cheers,
James (closet Nord fan)
x

My pleasure!

I hope the Bosie pans out for you Steve!

Why does this make me smile ... the effect is visible and audible

I don't have golden ears, so I often have to go off visual phase cues to know whether or not pedal down sympathetic resonance is implemented in a particular DP. Too much information?

Listening plus seeing is an amazingly powerful combo. You ought to try it some time - the Adobe Audition demo is fully functional.

Listening plus seeing is an amazingly powerful combo.

That might be, but the auditions for major symphony orchestras are held with the player behind a screen so the judges cannot be influenced by any visual movement; it just comes down to listening.

I always liked that.

There's a difference blind testing and forensic analysis. With the former you are trying to eliminate unnecessary bias, with the latter you are using every tool at your disposal to figure out what is going on.

Doing blind testing among DPs or between DPs and real pianos is valid, but doesn't interest me much because most PC pianos and some DPs are past the point of fooling me. I'd rather spend my time picking apart the inexact methods they are using to do this, as there is still considerable room for improvement.