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

My pleasure, and I'm happy that you find it useful. For me this has been a very fruitful group project, with others providing samples I never would have been able to obtain, and the analysis results shared with other interested parties.

I added the CP33 to the begging list of the DPBSD OP.

Thanks! Yes, I think it deserves a brownie point

Hi, Dewster, could this test go to the next version of DPBSD? (I can't see if it is already there implicitly, it would be a very basic direct physical behaviour to check).

Single Key Damper Check



(The next proposed testcase there I couldn't really understand):

That's test number 3, the "silent replay" test. I do the test with a single note.


That's test number 2, the "key sympathetic resonance" test. I do this test with notes both above and below the stimulus note.


I don't test for key sympathetic resonance with the pedal down because something that subtle sounding would be swamped by all the other undamped strings doing their thing (in a real piano). It would be a nightmare to try to hear something like that and report it with any certainty.

Or am I perhaps reading the test description incorrectly? Is the pedal depressed, the keys silently depressed, then the pedal lifted before the stimulus note is played?

WippenJackSpring? Beuller?

Dewster, I believe the proposed added tests actually are not in your original test... you test for their individual components, but not the 'complete' scenario all in one place.

The first proposed test is a single long-procedure test:
- Press a key + hold the pedal + silently replay that key + lift the pedal (without releasing the replayed key).

The silently replayed key should keep on sounding as the damper for those strings was always off the strings, it was never lowered... but, according to Temperament, most DPs, upon release of the pedal will quiet all keys even the ones that were depressed after.

The second procedure is:
- hold pedal + silently play a note + play a sympathetic note an octave or so higher.

The claim is that, there should be string resonance with the silently depressed key, but that, as you pointed out, is, IMHO, wrong. If the sustain pedal is depressed, what keys are silently held down is irrelevant... all dampers are always up and all strings with harmonics with the newly played note will resonate if the sustain pedal is depressed, so I really don't think that second part of the test is really accurate.

I hope I got the explanation right though

Rafa.

I believe you are describing the DPBSD silent replay test #3 (from the DPBSD readme file):

3. [C2v100t0]; [DPv127t1]; [C2v0t2]; [C2v1t3]; [DPv0t4]; [C2v0] & [G5v15] @ t5.

- Note C2 is played at velocity 100 at time 0 seconds.
- Damper pedal is fully depressed at time 1 second.
- Note C2 is released at time 2 seconds.
- Note C2 is played at velocity 1 at time 3 seconds.
- Damper pedal is fully released at time 4 seconds.
- Note C2 is released at time 5 seconds (and note G5 is played at velocity 15 to signal the end of the test.

Lots of DPs fail the penultimate step (note damps @ damper pedal release) which isn't the end of the world, but it does give some insight into how much internal bookkeeping they're doing with the keys and pedals.

I stand corrected... that is indeed the full test

Thanks for the clarification,
Rafa.

Thank You Dewster, at first reading I couldn't see it either that it is already included (therefore posted).

1.In the meantime I re-assessed the importance of this as a buying argument for a DP with the same conclusion as you have written:
2.
Had the same thought therefore couldn't understand it either.
3.
This would be a real test case for the completeness of SR implementation (as my CA51 would fail it). The problem probably is that no key sympathetic resonance will be activated while pedal down. (Once activated however it wan't get destroyed by pressing-lifting the sustain pedal, as I can reproduce it.)

And this behaviour could have more impact on the sound in real playing situations than the original reported cases! (While the player plays the keys he operates the sustain pedal independently).

Attila

Hi,

I made a recording with my brand new Casio AP 450.
The noise level is dreadful, but I recorded with my laptop through the headphones jack...

I asked Mike Martin if there is a way to make the DP record the midi file in wav format (my attempts where unsuccessfull)

I do not know if the current file is usable for DPBSD. If not I will try again this week-end.

I was unable to make it visible in the DPBSD media fire, so here is the link :
http://www.mediafire.com/?9ah55op7w5ryq39

Thanks! There is a rhythmic buzzing around -45dB that is pretty obnoxious, but it's in stereo and the peak levels are good.

Looking at the manual, I'm not certain but it seems like you might be able to do simultaneous MIDI playback and WAV recording directly on the AP450.

I must say the manual is not easy to understand ... the more so because there is no screen, and buttons are used for several functions
I was able to make a wav record of a midi record of something I played, but was unable to make the DP accept the .mid file as a recorded song (it is a .DSC file, and it does not look like the .CM2 format of casio midi ...)

After unsuccessfully trying the mid to wav conversion (I learned the midi file format in the process so it's not lost time) I made a new recording on my main computer. Many plug conversions were necessary (I only had RCA jack with enough cable length) but finally it's done. The result is here : dpbsd_v2.0_casio_ap450-2.mp3

Bye,

The noise floor is much better, but for some reason the stretch test is missing?

ooops, which test is it ? did I cut It at the beginning or end ? I'm making it ... again

Indeed, I must have accidentaly cut a part of the file. Here it is again : dpbsd_v2.0_casio_ap450-3.mp3

Hope it is ok, this time

Dewster, I have a proposition.

When I think about all these pianos, sometimes I get lost which one has which technology, and also sometimes when I look at sample lengths I forget what other pianos had. Also some people here may not have such technical mind to see some things as clear as you do. So my proposition is to make some kind of ranking or tech tree for pianos.

Pianos could have 0-10 note for technical aspects such as loop lengths, stretching amount, velocity layers, velocity blending quality, pedal effect and some of most important tests, eventually frequency and timbre balance (to point out pianos with weird timbre issues). 10 is no looping, 8 velocity layers etc. It could sum up to overall TECHNICAL pianos ranking or tech tree, which will of course not mean that playing experience will give you same conclusions. But it will give us and novice buyers nice and quite objective overview.

You already did the judgment, but it's verbal and in this way hard to gather in one place, like table, tree or ranking; also less understandable than numbers.

Other thing, much more controversial, would be adding keyboard quality note: for progressiveness, weight, sensor numbers. I'm not sure is this good idea, since it may start keyboard discussion war.

Looks good, thanks!


I've been asked this via PM a couple of times lately, and would love to comply, but I can't see exactly how.

If the text reviews were more numeric it would make more sense, but they are rather comment driven without rigid categories beyond pro/con/other, and Excel isn't the best at that kind of thing.

Many tests are pass / fail, stretching could be a ratio, attacks and decays are numeric. But there are too many caveats / qualifiers / things to say about almost everything. People might ask for (or I might be tempted to include) some math on the numbers for a grand score, but DPs are such balls of wax I'm kind of against one number summing them all up (rather like IQ, which is absurd on its face).

The DPBSD is really more of a no-go test for a DP you are already seriously considering buying, and not so much a pre shopping weeder. After reading the review you should download the MP3 and critically listen to it (particularly the looping, stretching, and layering tests) so you can make your own esthetic judgements (i.e. can you likely tolerate long term the way they've processed the sample set).

Dewster, there in in this topic about hammer delay we are discussing Casio's highly advertised new(?) hammer delay simulation. I can't see Hammer Delay was directly explored by DPBSD tests so far, but they could possibly significant information about DP quality. It could even help to differentiate between DP component quality separating keyboard and sound synthesis.

To determine is simply the time difference between MIDI Note ON Event and actual sound Attack time.

Time offsets between ppp and fff attack time of the same note: after Casio, there should be a consequent increasing delay from ppp to fff. Measuring Attack Delay for each single note should produce an increasing series of delay values for ppp,pp,p,mf,f,ff,fff. For each of these velocities should be a consequent delay series over all of 88 keys. Ideally, it should give a decreasing value from Low to High notes within one velocity.

If this kind of consequency check failes, it could reveal a contributor to overall jitter, which effects not only playability but even rendering too!

If this Hammer Delay behaviour is not measureable for a DP (ideally overall constant (low) delay values then) it could still be implemented within it's keyboard module and the delayed time values are delivered then to both internal sound synthesis module input and MIDI output.

Time Delay measurement could give important informations especially about SW Instrument: if they include such a Hammer Delay simulation, the MIDI keyboard input should not have it or otherwise it would result in a doubled effect.(It should then be possible to turn effect off at least on one side).

Time Delay Consistency Analysis could include:
Same note should produce constant timing for the same MIDI event.
Effect of repetitions to delay: should not have any impact on delay.
Playing a tone within a chord: should not have any impact on delay.

You have just the AP450 MP3s - do You see it possible to apply some of these measurements with current MIDI? (Or perhaps Yo have some of them built in already I have just overlooked). Thanx. Attila

I think you can change a variable related to hammer delay on some Roland keyboards? FP7F comes to mind.

Yamaha P35



Just in time for the Xmas season, PW forum member "1John" has supplied us with DPBSD MP3 of the new Yamaha P35! The MP3 (and pix) are of the default piano voice "Grand Piano 1". Those interested can also listen to the compressed layer test:

PIX: http://www.mediafire.com/?6blnhmfvd62bhoe
MP3: http://www.mediafire.com/?6s08mtkzmgz6rzw
MP3: http://www.mediafire.com/?jvo25kghymq0pjx (100:1 compressed layer test)

1John has this to say about the P35:


Analysis reveals fairly typical Yamaha sound technology for this price range. No pedal or key sympathetic resonance; attacks are short; loops are short and quivery / bland sounding and lacking any realistic interbeating or "wobble"; pretty stretched with 29 samples covering 88 notes; stretch group transitions are audible over most of the range.

Looking at the pan and phase views of the layer test doesn't indicate any velocity switching, but the uniform waveform view of this test reveals that this is a single layer sample set. Which doesn't mean there isn't significant timbre variation with velocity, there is over much of the range and it sounds fairly realistic, but it could perhaps be distributed more realistically because the highest 1/3 of velocity range doesn't have much variation. Overall the P35 seems pretty similar to the P85/95 but with somewhat longer decay lengths.

Poking through my past analysis pix archives, the P35 appears to share the same base sample session as the ~$800 USD Yamaha DGX-640. The P35 is currently selling for ~$450, for ~$150 more you could get the P-105 which has multiple layers, pedal sympathetic resonance, and better speakers, though the same GHS keyboard action and the same amount of stretching. Also in the $600 category is the Casio PX-150, which has pedal sympathetic resonance, 4 layers of longer samples, somewhat less stretching, and arguably better keys, though the speaker system is perhaps not as good as the P-105. The low end is kind of a tough place for shoppers (truth be told the mid and high ends aren't much easier).

Muchas gracias to 1John for this timely DPBSD MP3!

Some analysis pix and text review:


Figure 1. Spectral pan view of the pedal sympathetic resonance test, pedal down @ left, pedal up @ right, stimulus removed, normalized to -1dB. No pedal sympathetic resonance visible or audible.


Figure 2. Waveform view of the entire looping test, vertical zoom applied to see the noise floor. Note decay is long over the low end, rather short over the rest of the range.


Figure 3. Spectral phase view of the layer test, highly compressed. No visible or audible layer switches. You can listen to this as a separate MP3 file (listed above).


Figure 4. Spectral frequency view of the layer test. There is significant timbre variation, but not a lot at the upper end of velocity where you might expect it.


Figure 5. Waveform view of the layer test. Uniformity of the waveform throughout the test indicates a single layer is employed.



Figures 6a & 6b. Spectral phase view of the stretch test, mid notes, normalized to -1dB to increase clarity, P35 top, DGX-640 bottom. Sonic "fingerprint" match indicates these two models share the same sample session. 29 stretch groups are clearly visible and audible over most of the range.


--------------
- Yamaha P35 -
--------------
FILE & SETUP:
- dpbsd_v2.0_yamaha_p35.mp3
- yamaha_p35_layers_comp.mp3
- Recorded using MusE, Lexicon Lambda, LAME encode to MP3.
- This is the first patch: "Grand Piano 1".
- Recorded by "1John".
PROS:
- Passes the silent replay test.
- Passes the quick partial damping test.
- Passes the late pedal partial damping test up to the test limit of 0.5 seconds.
- Passes the half pedaling test.
- Low note decays are fairly long (~Pianoteq), mids and highs could be longer (~1/2 Pianoteq).
CONS:
- Fails the pedal sympathetic resonance test.
- Fails the key sympathetic resonance test.
- Partially damped notes don't sound "buzzy".
- Obviously looped, the loops sound static rather than "wobbly".
- Attack sample lengths are (C1:C8): 2.2,2.2,1.7,1.6,1.3,1.1,?,? seconds.
- Loop sample lengths are (C1:C8): 0.7,0.7,0.5,0.5,?,0.3,?,? seconds.
- Stretching is visible over the entire range, audible over the low and mid notes.
- Stretch distances: 2,3(x27),5 = 29 groups.
- No obvious key up or pedal down sound effects (via MIDI).
- I believe this is a single velocity layer sample set.
- Very little timbre variation in top 1/3 of velocity range.
OTHER:
- Phase "fingerprint" same sample session as DGX-640.
- Notes played @ vel=1 produce no sound.
- Dynamic range ~65dB (vel=1:127).
- Dampered | undampered transition: F#6 | G6.
- L&R are likely swapped in the recording.
- MP3 levels: peak @ -3.8dB, noise floor @ -80dB.
- Date reviewed: 2012-11-25