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

Dewster, thank you for your new test "5. Late pedal partial damping test"! When I made my suggestions for new tests in This Post earlier on page 54, I suggested you two tests. You agreed then and implemented my second suggestion as "4. Quick partial damping test". I am glad to see that my first suggestion goes now into this new "5. Late pedal partial damping test" Thank you again.

May I remind you of yet another suggestion I made in This Post on page 64 which concerns pressing the damper pedal after keys have been pressed and held in which case you should hear the resonance sound engaging. Most pianos and software libraries activate resonance samples only when notes have been played after the damper pedal have been pressed but not the reverse. Thank you in advance and sorry if you already have that test (I checked through you technical descriptions but wasn't able to recognize it).

Gosh, that was so long ago I don't remember that you suggested it CyberGene! Thanks very much for your suggestions, maybe my subconscious didn't forget?


I can sometimes hear this effect at the beginning of the silent replay test, where the C2 note is played at velocity 100 and held, and the damper pedal is pressed one second later.

When I was changing the third section of the sympathetic resonance test in v2.0 I considered doing a pedal pump rather than just the lift, but I found it rather difficult to hear.

Indeed! Very good, so we can only re-review the existing tests to see if they support that "damper resonance reengaging" or whatever we can call it best. In fact, I remember that was heavily advertised first by Steinberg The Grand, referred by Steinberg to as "repedaling" but I think most pianists are using the term repedaling differently (referring rather to "5. Late pedal partial damping test" I suppose).

Ohh, and don't worry about the fact you've missed my suggestion. I know the thread is really huge. Also, test 4 is covering to some degree the main functionality.

Also, I may have argued with you sometimes regarding the DPBSD test but it's not the test itself that have bothered me since the scientist in me (if a computer programmer can be considered scientist) really loves the scientific, exact and thorough approach. If I have had any objections in regards to DPBSD, that has been mostly to the conclusions drawn by the results and not the test itself. I hope I haven't offended you with such arguments. Anyway, I would be glad to contribute to the test evolution and improvement

I obliquely noted this in today's addition to the NX review above:

- Passes the silent replay test, pedal sympathetic resonance can be heard during it.

The NX is pretty remarkable in terms of replicating AP key/pedal behavior.


I've struggled with terminology throughout this project, and finally reached the point where I decided to call things what I though would be clearest, even if it wasn't the "accepted" norm. For instance, too many things are called "resonance" IMO.


I've also struggled with the objective / subjective thing and you (and others here) should definitely feel free to object if you disagree with anything I write - I have learned an enormous amount from the very knowledgeable members here at PW. With all of this unintentional ear training, I've perhaps become too comfortable with including a fair amount of subjective content in my reviews.

I'll entertain any and all suggestions to improve the DPBSD. Thank you very much for yours CyberGene, I'm sorry I forgot and therefore neglected to credit you as the origin of these latest test additions.

RD-700NX MIDI Playback

This post is devoted to getting the most out of the RD-700NX SN piano voices when playing them via MIDI rather than via the built-in keyboard.

Some here are painfully aware that most (all?) Roland SN DPs have issues with incidental piano effects and sounds (sympathetic resonance, pedal down "loom of strings", etc.) going missing when playing a MIDI file. Often these can be turned back via MIDI SYSEX commands, but it can be a tough nut to crack. For instance, here is a web page made by another PW forum member which is explains how to turn things back on in the HP-307 (before any MIDI implementation doc from Roland):

http://wmsar.info/roland/

What's weird about the NX, compared to the GX/F, is that Roland yanked the pedal sympathetic resonance in-line effect from the generic MFX pool. In the GX/F I presume it can be freely used as an effect with any voice (though in the first MFX slot only) whereas in the NX pedal sympathetic resonance is seemingly more tightly bound and only available to the SN acoustic pianos. After this change of course Roland neglects to tell us where the sympathetic resonance effect exists in MIDI SYSEX address space, so those of us who want to manipulate its parameters via MIDI are SOL unless we stumble upon them either by accident or through experimentation or until Roland tells us in an updated document.


Anyway, here are my experiences with the NX with MIDI playback so far:

1. I unplugged the NX and schlepped all of its daggone 25kg / 55lbs downstairs to my PC.

2. To clear the air, I performed a factory reset on the NX ( Menu | Utility | Factory Reset All ).

3. I found that if I played the DPBSD MIDI file (which targets MIDI channel 1 with no bank or program changes) via Sonar 6 I got what sounds like a non-SN piano, probably from the GM bank, which is looped, has a relatively short note decay time, is heavily stretched, and has a fairly jarring layer switch. It seemed no amount of fiddling with the front panel controls would switch the MIDI voice or alter it in any way.

4. I changed the "System Part" mode ( Menu | System ) from "16PART+PERF" to "16PART" and performed a write. Now I found I could change the voice (piano, clav, guitar, etc.) via the front panel and the MIDI playback would follow that. Next I tried changing the various SN piano settings, and I found most were clearly present during MIDI playback while a couple of others weren't:

Editable / Present:
- Base Voice selection
- Stereo Width
- Nuance (subtle stereo phase tweak)
- Lid (how open the virtual lid is)
- Duplex Scale (resonant strings / bars in some pianos)
- String Resonance (key sympathetic resonance)
- Key Off Resonance (string damp buzz)
- Hammer Noise ("knock")
- Tone Character (changes basic sound, roughly dark to bright, but more complex than that)
- Sound Lift (MIDI velocity shift & scale)
- Micro Tune Edit (menu, individual offset tuning per note)
- Sym.Resonance (menu, pedal sympathetic resonance)
- Equalizer (menu, 4 band equalizer)

Not Editable / Present:
- Damper Noise ("loom of strings")
- Key Touch Edit (MIDI velocity shift & scale)

I guess I don't care all that much if the Key Touch Edit doesn't alter anything during MIDI playback, though it would have been nice - I can edit the MIDI itself to accomplish this I suppose. But the fact that the "loom of strings" pedal down sound is missing is kind of strange. Why, of all things, did they leave that out? Is it because the loudness of the effect is based on pedal velocity, and they were afraid the default level would sound too loud if people ran random MIDI files through it that had only pedal on/off (0, 127)? Even if that's the case, turning the damper noise off for MIDI means MIDI recorded and played back on the NX won't sound the same as playing it via the built-in keyboard, which seems like a worse thing to me. Who knows.

I found that running the DPBSD file would put a '*' next to the selected voice name to indicate that it had been edited. This is most likely due to the fact that the DPBSD zeros out the reverb send level at the beginning of the file.


SYSEX TEST 1 - I was able to send the NX a MIDI SYSEX "identity request" message and receive a response. I did this in Sonar 6 by making SYSEX bank 0 defined as the ID request below. In a MIDI track I insterted a SYSEX bank 0 send event. I then made another MIDI track and recorded it while playing the first track. Kind of round about, but it seems to do the trick. I got exactly what the manual said I would as a response.

TX: F0 7E 10 06 01 F7
RX: F0 7E 10 06 02 41 50 02 00 00 00 01 00 00 F7

TX decoded:
F0 - SYSEX start
7E - universal non-realtime indicator
10 - target device ID (Roland says use 10 thru 1F)
06 - general info category
01 - ID request
F7 - SYSEX end

RX decoded:
F0 - SYSEX start
7E - universal non-realtime indicator
10 - target device ID
06 - general info category
02 - ID rply
41 - ID # indicating this is a Roland
50 02 - device family code for RD-700NX
00 00 - device family number code for RD-700NX
00 01 00 00 - software revision
F7 - SYSEX end


SYSEX TEST 2 - Next I sent example 1 in the "How to calculate the checksum" section of the NX MIDI implementation manual, which is a SYSEX data set command:

TX: F0 41 10 00 00 50 12 10 00 04 00 02 6A F7

And I saw the front panel CHORUS/DELAY LEDs switch from CHORUS to DELAY. Note that there is an error in the manual here, they left out one of the "00" groups in the "Model ID" field at the end of the example.


Next I sent a command to switchg CHORUS/DELAY back to CHORUS:

TX: F0 41 10 00 00 50 12 10 00 04 00 01 6B F7

And I saw the LEDs switch back to CHORUS. I calculated the checksum with a nice little program I found on the web called "Roland Checksum Calculator" which is a small Windows executible made and distributed by some kind soul.


SYSEX TEST 3 - Next I sent example 2 in the "How to calculate the checksum" section of the NX MIDI implementation manual, which is a SYSEX data request command:

TX: F0 41 10 00 00 50 11 10 00 00 00 00 07 03 0B 5B F7
RX: - no response -

I checked the checksum and 5B appears to be correct, so I don't know what the heck's going on here.


SYSEX TEST 4 - Next I tried setting the "Damper Noise Level" @ address 0x10020005 to the max of 0x7F:

TX: F0 41 10 00 00 50 12 10 02 00 05 7F 6A F7

And when I was in the "TONE EDIT" screen for the first SN piano I saw the "Damper Noise Level" number jump from 29 to 127. Success it would seem. But I still don't hear the damper pedal down noise in the DPBSD so I won a minor battle but am still losing the war.


SYSEX TEST 5 - Next I tried reading the "Damper Noise Level" @ address 0x10020005:

TX: F0 41 10 00 00 50 11 10 02 00 05 00 00 00 01 68 F7
RX: - no response -

I'm clueless as to why data requests don't produce a response. I guess I should put an LED on the MIDI port to see if it's trying to tell me anything that might be being filtered out somehow by Sonar.


--------------
- 2011-02-11 -
--------------
Here's my email to Roland support, with the title "Damper noise missing during MIDI playback":




--------------
- 2011-04-02 -
--------------

The latest NX software load from Roland (v 1.03) pretty much fixed the note decay chopping and the layered string dropping issues - yay! Roland also mentions something about it fixing MIDI SYSEX data requests as well:


So today I thought I'd give SYSEX another try.

First I tried the SYSEX "identity request" message as I did previously and I again received a response:

TX: F0 7E 10 06 01 F7
RX: F0 7E 10 06 02 41 50 02 00 00 00 01 00 00 F7

Strange that the software version number doesn't change? Then again, I had 1.02 in there from the factory and it reported the same "00 01 00 00" as well. Whatever.

Next I again tried example 2 from the "How to calculate the checksum" section of the NX MIDI implementation manual, which is a SYSEX data request command (data dump):

TX: F0 41 10 00 00 50 11 10 00 00 00 00 07 03 0B 5B F7
RX: F0 41 10 00 00 50 12 10 00 00 00 43 6F 6E 63 65 72 74...

Tons of data in multiple SYSEX messages came back so this is fixed by v1.03 as well! This data is the current status of the "Live Set (temporary)" area, which seems to represent the current settings of / edits to the currently selected patch. Switching the patch via the front panel button will make any edits disappear, so you need to explicitly save them if you want them to stick.

Next, while thumbing through the "Live Set (temporary)" section of the MIDI implementation manual, I stumbled upon something called "Live Set Resonance" and wondered if this was referring to "pedal down sympathetic resonance". I was complaining above that this in-line effect was yanked from the MFX pool and that I couldn't find it in the implementation guide. Anyway, I made a quick SYSEX MIDI file to test this by turning it off:

TX: F0 41 10 00 00 50 12 10 00 20 00 00 50 F7

I went into the "TONE EDIT" menu, arrowed over and down to enter the "Sym.Resonance" menu, and watched the "Sw:" entry while I sent the above SYSEX. The entry went from "ON" to "OFF" - so I found the sympathetic resonance in SYSEX space (where no one can hear you scream)!

All my MIDI files are here if anyone wants to play with them:
http://www.mediafire.com/?9flwlzgltrbetkv

I also got a response from Roland support regarding the missing damper noise during MIDI playback. They want me to call them and discuss it but I haven't done so yet. When I do I'll report back here.

Interesting stuff.

I think you're at the stage now where you should contact Roland technical support (assuming you haven't already) requesting detailed information on how to implement all of the SN features using MIDI.

Given the great efforts that individuals such as yourself are going to to harness the instrument's sounds, I wouldn't be surprised if you receive a response from one of the Roland engineers directly. I know that if I received such a request from a Kawai customer I'd certainly pass it along to the engineering team and encourage them to respond.

Give it a try - you may be pleasantly surprised.

Cheers,
James
x

Speaking about that, James, have you had any response from engineering team regarding that same issue on Kawai CA-line of instruments?

CyberGene, I'm sorry but I don't know.

Kind regards,
James
x

Good suggestion James, I'll give that a go and report back here if/when they reply.

Casio Privia PX-330 Revisited



A fifth DPBSD MP3 submission from anotherscott, this time for a re-review of the Casio Privia series PX-330 - much thanks for the continued support of this project! The first DPBSD MP3 for this DP was v1.3 from pesk, and since several new tests have been introduced in the interim it's really great to have the opportunity to revisit this DP for a more in-depth analysis.

As usual for the DPBSD test, the MP3 is of the default voice "Grand Piano Modern". But as a bonus anotherscott also recorded the seventh voice "Grand Piano Classic" which I have also uploaded to the share point:

PIX: http://www.mediafire.com/?e537yfru2s73q29
MP3: http://www.mediafire.com/?17o4gdvjhtdb3cb
BONUS MP3: http://www.mediafire.com/?3j4l2w7iz7r514f (the 7th patch "Grand Piano Classic").

Listening to the seventh patch, it sounds highly similar to, but somewhat mellower than, the first main patch. Analysis shows that they indeed share the same sample set, but with what I believe likely is some non-linear velocity scaling and EQ going on with the seventh patch. Here are anotherscott's comments about these two piano voices:



The PX-330 is first and foremost a "value" piano available at many popular retail outlets, so tons of these units are undoubtedly sold. It helps that it checks a lot of the boxes often found only on more expensive DPs: weighted graded hammer action triple sensor keys, built-in speakers, music rest, pitchbend wheel, dedicated control buttons, DIN MIDI in/out, 1/4" line in/out, USB, LCD display, two headphone jacks, SD card slot, rhythms, GM sound set, 16 track recorder, light weight (25lbs), etc. From a purely price/feature-set perspective it's quite amazing.

From a sound technology perspective, the piano note decays are surprisingly long. But it is audibly looped, with fairly short attack samples and very short loop samples. The lower loops sound "wobbly" to me, which I think I prefer over more static sounding loops, though it would certainly benefit greatly from somewhat longer sample lengths. It is also highly stretched, with the 27 stretch group transitions fairly audible over the low and mid notes. I can see evidence of three velocity layers (Casio reports 4) though timbre change with velocity is very smooth with no audible switching or timbre steps.

I believe it has some kind of pedal sympathetic resonance, but the effect is so subtle I really can't hear it. There is no key sympathetic resonance that I can detect. A lot of what gives a real piano that rich sound IMO is sympathetic resonance, so to have these features weak & absent is rather unfortunate, though perhaps that's asking too much from a DP this inexpensive.

In terms of key / pedal interaction, it supports partial quick damping and half pedaling, but it fails the silent replay test with the test note damping at pedal up. Other than string damping, I can't hear any obvious pedal up/down or key up sound effects.

I run across the PX-330 all the time and so get many opportunities to give it a spin. The keys don't feel too bad and the piano sound is really pretty OK, particularly when price (and other features) are factored in. But the other voices in there are by-and-large pretty lame.

Again, much thanks to anotherscott for the two up-to-date test samples of this highly popular DP!

Some analysis pix and text review:


Figure 1. Spectral pan view of the pedal sympathetic resonance test, stimulus removed, normalized to -1dB to increase clarity, pedal down @ left, pedal up @ right. Something appears to be going on, but I unfortunately can't hear it.


Figure 2. Spectral frequency view of the key sympathetic resonance test. No visible or audible key sympathetic resonance.


Figure 3. Waveform view of the entire looping test, vertical zoom applied to see the noise floor. Note decay is nice and long.


Figure 4. Spectral phase view of the note C4, normalized to -1dB to increase clarity, cursor at attack / loop transition. Attack and loop sample lengths are rather short.


Figure 5. Spectral phase view of the stretch test, mid notes, normalized to -1dB to increase clarity. This DP is highly stretched, with fairly uniform stretch group size (mostly 3), and transitions are audible over the low and mid note ranges.


Figure 6. Spectral frequency view of the layer test. Timber variation with velocity is uniform and smooth.


Figure 7. Spectral pan view of the layer test, compressed 20:1. Two visible layer switches, the transitions of which are not audible to me (Casio claims 4 layers).


-----------------------
- Casio Privia PX-330 -
-----------------------
FILE & SETUP:
- dpbsd_v1.9_casio_px330.mp3
- Sequenced and recorded on Mac with ProTools LE, Mbox interface @ 24-bit, MP3 via Quicktime Pro.
- This is the first patch: "Grand Piano Modern".
- Recorded by "anotherscott".
PROS:
- Something visibly going on with pedal sympathetic resonance.
- Passes the quick partial damping test.
- Passes the half pedaling test.
- Note decays are nice and long.
- This is a smoothly blended multi-velocity layer sample set (Casio reports 4 layers, three are visible).
- Timbre change with velocity is nicely spread out with no audible timbre steps.
CONS:
- Pedal sympathetic resonance is so subtle I almost can't hear it.
- Partially damped notes don't sound "buzzy".
- No visible or audible key sympathetic resonance.
- Fails the silent replay test - note damps @ pedal up.
- Obviously looped, the loops sound "wobbly" rather than static.
- Attack sample lengths are (C1:C8): 1.8,1.5,1.4,1.4,1.4,1.3,1.0,1.0,? seconds.
- Loop sample lengths are (C1:C8): 1.6,1.1,1.0,0.8,1.0,1.0,0.9,?,? seconds
- Stretching is visible over the entire range, audible over the low and mid notes.
- Stretch distances: 3,2,3(x16),4,2,3(x5),4,5,5 = 28 groups.
- No obvious pedal up/down or key up sound effects.
OTHER:
- Notes played @ vel=1 produce no sound.
- Dynamic range ~49dB (vel=1:127).
- MP3 levels: peak @ -1.0dB, noise floor @ -81dB.
- Recorded with L & R swapped.
- Almost certainly identical to the first patch "Grand Piano 1" in the PX-3.
- Date reviewed: 2011-01-28

Casio Privia PX-3 Review



A sixth DPBSD MP3 submission from anotherscott! This MP3 is from the Casio Privia PX-3, the default main AP voice "Grand Piano 1". But as a bonus anotherscott also recorded the second voice "Grand Piano 2" which I have also uploaded to the share point:

PIX: http://www.mediafire.com/?87gqqihj4us9ldg
MP3: http://www.mediafire.com/?oeaqcto1tpz7zf9
BONUS MP3: http://www.mediafire.com/?rbubab0xgorxwal (the 2nd patch "Grand Piano 2").

My analysis and careful listening tell me that these two voices are almost certainly identical to the first and seventh piano voices (respectively) in the Privia PX-330, so I'll direct you to that post (LINK) for a review of the sound technology in the PX-3 rather than repeat myself here.

I ran across the PX-3 at our local Guitar Center a couple of months ago but I didn't play around with it enough to form any solid opinions. From a brief comparison of the manuals it looks like they share pretty much the same knobs, buttons, I/O, and number of tones and effects. The PX-3 lacks the speakers found on the PX-330 but gains an "ivory touch" feel to the keys. Surprisingly the MIDI recorder function is missing in the PX-3, though the ability to play MIDI files is still there. Duet mode and the metronome are gone, with the corresponding buttons given over assignable functions. Also missing are the auto-accompaniment and rhythms of the PX-330. The PX-3 is probably the better choice between the two if you are looking for a portable MIDI controller as it has four assignable zones - and indeed this is the way it is marketed.

Now over to anotherscott who owns the PX-3 and has many insightful comments regarding it:



Thanks again to anotherscott for providing the analysis files and in-depth comments for this inexpensive stage piano / controller!

Text review:

---------------------
- Casio Privia PX-3 -
---------------------
FILE & SETUP:
- dpbsd_v1.9_casio_px3.mp3
- Sequenced and recorded on Mac with ProTools LE, Mbox interface @ 24-bit, MP3 via Quicktime Pro.
- This is the first patch: "Grand Piano 1".
- Recorded by "anotherscott".
PROS:
- Something visibly going on with pedal sympathetic resonance.
- Passes the quick partial damping test.
- Passes the half pedaling test.
- Note decays are nice and long.
- This is a smoothly blended multi-velocity layer sample set (Casio reports 4 layers, three are visible).
- Timbre change with velocity is nicely spread out with no audible timbre steps.
CONS:
- Pedal sympathetic resonance is so subtle I almost can't hear it.
- Partially damped notes don't sound "buzzy".
- No visible or audible key sympathetic resonance.
- Fails the silent replay test - note damps @ pedal up.
- Obviously looped, the loops sound "wobbly" rather than static.
- Attack sample lengths are (C1:C8): 1.8,1.5,1.4,1.4,1.4,1.3,1.0,1.0,? seconds.
- Loop sample lengths are (C1:C8): 1.6,1.1,1.0,0.8,1.0,1.0,0.9,?,? seconds
- Stretching is visible over the entire range, audible over the low and mid notes.
- Stretch distances: 3,2,3(x16),4,2,3(x5),4,5,5 = 28 groups.
- No obvious pedal up/down or key up sound effects.
OTHER:
- Notes played @ vel=1 produce no sound.
- Dynamic range ~49dB (vel=1:127).
- MP3 levels: peak @ -3.3dB, noise floor @ -81dB.
- Almost certainly identical to the first patch "Grand Piano Modern" in the PX-330.
- Date reviewed: 2011-01-28

Kawai MP6 Review



Thanks to Piano World forum member "JFP" we have a DPBSD MP3 of the Kawai MP6! This is the fifth Kawai DP reviewied in this thread.

- DPBSD MP3 & analysis pics -
MP3: http://www.mediafire.com/?ve5fq87lwglf4k5 (DPBSD MP3)
MP3: http://www.mediafire.com/?6dmc9ry9gr8b2s1 (key sympathetic resonance)
MP3: http://www.mediafire.com/?69yrrg5urtdr3yp (MP10/MP6 layer mashup)
MP3: http://www.mediafire.com/?hpw6ctwmvbebkyp (compressed layers)
PIX: http://www.mediafire.com/?idxvt84ct0443yl

The MP6 pedal sympathetic resonance effect is realistically subtle, and does a lot to smooth out the quick pulsating sound of a group of decaying notes looping together. The key sympathetic resonance also sounds nice, with both lower and higher notes subtly responding - unfortunately it doesn't seem to work when playing the piano voice via MIDI. JFP kindly suppled a separate MP3 where the keyboard is manually played so you can hear this effect for yourself.

JFP also confirmed that there is a key-up sound effect associated with playing lower notes loudly and then releasing them quickly. It isn't clear to me as to whether or not this plays back via MIDI, and the DPBSD don't have a specific test for it that might make it more audible.

The sample set of the main piano seems to be some kind of blend between CN33 (PHI) and MP10 (UPHI). The MP6 has the same visible (though not audible in either DP) velocity layer switches of the CN33, but the lowest notes of the MP6 have longer attack and decay samples than the CN33. In fact, as JFP pointed out to me, the attack and decay sample lengths are a fairly close match to the MP10, which is interesting:

- Visible layer switching -
CN33: vel=86,102,112
MP6 : vel=86,102,112

- Attacks sample lengths (C1:C8) -
CN33: 1.8,2.1,2.1,1.8,1.7,1.6,1.1,? seconds.
MP6 : 2.4,2.7,2.1,1.5,1.7,1.5,1.1,? seconds.
MP10: 2.5,3.0,2.2,1.8,1.7,1.5,1.1,? seconds.

- Loop sample lengths (C1:C8) -
CN33: 0.8,0.8,0.7, 0.8, 0.7, 0.7, 0.6,? seconds.
MP6 : 1.5,1.3,0.7, 0.8, 0.65,0.75,0.65,? seconds.
MP10: 1.4,1.3,0.66,0.83,0.65,0.75,0.65,? seconds.

I did a mashup of the MP10 and MP6 layer test hoping to better hear any differences between them (link above). I don't hear tons of difference, though the MP10 sounds a bit more "open" than the MP6. The brightest layer at the top end of the MP10 velocity has a fairly jarring layer switch:

- MP10 / MP6 layer mashup -
1a. MP10 C4 velocities 1,3,5,7,9,11,13,15
1b. MP6 C4 velocities 1,3,5,7,9,11,13,15
1c. 0.5 sec mute
2a. MP10 C4 velocities 17,19,21,23,25,27,29,31
2b. MP6 C4 velocities 17,19,21,23,25,27,29,31
2c. 0.5 sec mute
...
8a. MP10 C4 velocities 113,115,117,119,121,123,125,127
8b. MP6 C4 velocities 113,115,117,119,121,123,125,127

As usual I compressed the layer test of the MP6 20:1 in Audition to listen to timbre variation with velocity while largely negating amplitude changes through the test. This also helps to make the layers more visible in the spectral pan and phase views in Audition. I've including it here for anyone that wants to listen to it (link above) as well as a link to the similar file for the MP10. The upper ~1/4 velocity range of the MP6 sounds rather static to me. (Pay no attention to the buzzy mosquito noise at the beginning of each, that's an artifact from a previous normalization.)

Taking a quick peek at the manual, it appears the MP6 shares the no-nonsense controls and intuitive layout of the MP8, with added MIDI and audio recording functionality. The pitchbend and modulation wheels are located on the control panel rather than in the keybed area. No wall-wart, and it has a music rest. Pretty sweet.


Some analysis pix and text review:


Figure 1. Spectral pan view of the pedal sympathetic resonance test, stimulus removed, normalized to -1 dB to bring out detail, pedal down (left), pedal up (right). The effect sounds subtly pleasant to my ears and helps to break up audible looping.


Figure 2. Spectral frequency view of the key sympathetic resonance test played manually. First "blob" is the C4 stimulus plus resonance, later (at cursor) C3 is lifted, later C5 & C6 are lifted. The effect is audibly pleasant with both lower and higher notes subtly responding.


Figure 3. Waveform view of the looping test with vertical zoom applied. Decay times for the low and mid notes are nice and long.


Figure 4. Spectral phase view of the note C3, normalized to -1 dB to bring out detail, cursor at end of attack crossfade. Attack & loop samples are quite clearly seen here.


Figure 5. Spectral pan view of the stretch test, normalized to -1 dB to bring out detail, mid notes. No visible or audible stretching.


Figure 6. Spectral pan view of the layer test compressed 20:1. Three layer switches are visible, but not audible (they are smoothly blended).


Figure 7. Spectral frequency view of the layer test. Timber variation is smooth over most of the velocity range, though rather static sounding in the top ~1/4 range.


-------------
- Kawai MP6 -
-------------
FILE & SETUP:
- dpbsd_v2.0_kawai_mp6.mp3
- This is the main "Concert 1" piano voice.
- MacBook Pro / Logic Pro 9.x => USB MIDI => MP6 => MP3 recorded directly to USB stick.
- Recorded by "JFP".
PROS:
- Passes the pedal sympathetic resonance test, the effect is pleasant sounding but subtle.
- Passes the key sympathetic resonance test, strings both lower and higher subtly respond.
- Passes the quick damping test.
- Passes the half pedaling test.
- Passes the late pedal partial damping test up to the test limit of 0.5 seconds.
- Low note decay times are nice and long.
- No obvious stretching visible or audible.
- Timbre variation is fairly smooth with increasing velocity, layers are well blended.
- Velocity layer switch @ vel=86,102,112 visible but not audible.
CONS:
- Key sympathetic resonance is missing when playing via MIDI.
- Fails the silent replay test, note damps @ pedal up.
- Decay times are a bit short with the higher notes.
- Attack sample lengths are (C1:C8): 2.4,2.7,2.1,1.5,1.7,1.5,1.1,? seconds.
- Loop sample lengths are (C1:C8): 1.5,1.3,0.7,0.8,0.65,0.75,0.65,? seconds.
- Visibly looped over entire range, audibly looped over lower-mid to mid note range.
- Not a lot of timbre variation at the highest velocities (~top 1/4 range).
- No pedal up/down sound effects.
OTHER:
- Notes played @ vel=1 produce a sound (this is an editable option).
- Dampered | undampered transition: F#6 | G6
- Dynamic range 44.7dB (vel=1:127).
- MP3 levels: peak @ -4.8dB, noise floor @ -84dB (very approx.).
- Date reviewed: 2011-02-08.

Kawai MP10 Piano Voices Compared

I find it really interesting when DP manufacturers go to the trouble of including fundamentally different piano samples in a particular model. The main piano patch is the usually best, with the other piano voices rather more compromised in terms of implementation, at which point one is perhaps left wondering how much better the main voice could have be had they dispensed with the secondary voices altogether and used the freed-up resources for the main piano. The flip side of this coin is if the other voice or voices are good enough, then the player has a fallback position should the main piano be faulty, not to one's taste, or not suited to a particular playing style.

Much like the HI / PHI / UPHI mashups I did a while ago, I thought it might be interesting to listen to a few mashups of three different acoustic piano voices in the latest Kawai MP10, which are "Concert Grand", "Pop Piano", and "Jazz Grand 1". Here is how Kawai characterizes them in the MP10 manual:

• Concert Grand - A rich and dynamic concert grand piano.
• Pop Piano - A clear and vibrant pop grand piano.
• Jazz Grand 1 - A warm, powerful grand piano sound with a vintage jazz character.

I. Attacks & Loops


Figure 1. Spectral phase view of the notes C3, C4, and C5 in the first mashup. Attack and loop sample lengths are roughly comparable across sample sets for these three notes.

MP3: http://www.mediafire.com/?4dk02ygh8y0hmr3

The first mashup consists of the notes C1, C2, ..., C7, C8 as follows:
C1: Concert, Pop, Jazz (9 seconds each)
C2: Concert, Pop, Jazz (8 seconds each)
C3: Concert, Pop, Jazz (7 seconds each)
C4: Concert, Pop, Jazz (6 seconds each)
C5: Concert, Pop, Jazz (5 seconds each)
C6: Concert, Pop, Jazz (4 seconds each)
C7: Concert, Pop, Jazz (3 seconds each)
C8: Concert, Pop, Jazz (2 seconds each)

I limited the sample lengths so that the loop sections are adequately represented but not so long as to induce ennui in the listener. Each DP sample set was bulk normalized to -1dB peak before being mashed.

What I hear:
- C1: Concert is least bright (though it is still a bright piano sound), Jazz is least wobbly
- C2: All 3 sound OK.
- C3: Concert decay sounds best, Pop decays too quickly, Jazz decay tone kind of strange.
- C4 & C5: Concert and Pop sound very similar, Jazz has a nice phasey sound.
- C6: All 3 sound very similar.
- C7: Concert & Pop sound very similar, Jazz has more sympathetic resonance.
- C8: All 3 sound very similar.


II. Layer Dynamics


Figure 2. Waveform view of the entire second mashup.

MP3: http://www.mediafire.com/?8w9pqclccnnnfyf

The second mashup consists of the note C4 (all durations 0.5 sec) as follows:
Concert @ velocity 1, 3, 5, 7; Pop @ velocity 1, 3, 5, 7; Jazz @ velocity 1, 3, 5, 7
Concert @ velocity 31, 33, 35, 37; Pop @ velocity 31, 33, 35, 37; Jazz @ velocity 31, 33, 35, 37
Concert @ velocity 61, 63, 65, 67; Pop @ velocity 61, 63, 65, 67; Jazz @ velocity 61, 63, 65, 67
Concert @ velocity 91, 93, 95, 97; Pop @ velocity 91, 93, 95, 97; Jazz @ velocity 91, 93, 95, 97
Concert @ velocity 121, 123, 125, 127; Pop @ velocity 121, 123, 125, 127; Jazz @ velocity 121, 123, 125, 127

Each DP sample set (velocity 1:127) was bulk normalized so that velocity 127 = -1dB peak before being mashed.

What I hear:
- Concert sounds nice and bright, Pop sounds more nasal (forward?) than the others.


III. Layer Timbre Variation


Figure 3. Spectral frequency view of the third mashup. All three voices seem to have a comparable number of layers. Layer transitions are most visible for the Concert and Jazz pianos, the Pop piano layer transitions appear more blended.


Figure 4. Spectral pan view of the third mashup. Layer transitions are visible for all. Jazz stereo image is rather erratic looking.

MP3: http://www.mediafire.com/?htdkdfo63rdhayn

The third mashup consists of the note C4 (all durations 0.5 sec) as follows:
Concert @ velocity 1, 3, 5, ..., 123, 125, 127
Pop @ velocity 1, 3, 5, ..., 123, 125, 127
Jazz @ velocity 1, 3, 5, ..., 123, 125, 127

Each sample set (velocity 1:127) was bulk compressed 20:1 to largely eliminate any amplitude variation but leave in place any timbre variation.

What I hear:
- Concert: Several of the early layer switches are audible, the very last one is quite audible.
- Pop: The best blended of the three.
- Jazz: Severe stereo image and timbre issues between several of the layers.


Conclusions
The Concert voice could have better blending of the layers, particularly the transition to the highest one, but is otherwise nicely done. The Pop voice seems to be around the same quality level as the Concert, with better layer blending. The Jazz voice is perhaps unusable due to the severely mismatched layers.

I'd be very interested in the opinions of PW members who own and play the MP10 as to the character / similarities / differences / likes / dislikes of the various piano voices in the MP10.

Yamaha Motif Rack XS Review



Another DPBSD MP3 submission from anotherscott (the seventh in fact) this time for the Yamaha Motif Rack XS "Full Concert Grand" main piano patch.

MP3: http://www.mediafire.com/?e92ac2oeheo3ceu
PIX: http://www.mediafire.com/?0p7wuewty6ed971

The XS piano voice is rather typical Yamaha, with echoy sounding pedal sympathetic resonance, and highly processed rather static sounding loops. I was a little surprised by the fact that the layers aren't very blended, with most timbre variation happening at the layer transitions. I also noticed at the start and end of the half pedaling test that partial pedaling modulates the note decay time but not the pedal sympathetic resonance effect, which strikes me as rather unrealistic. The resonance effect seems to be on or off with no in between, yet the decay time of the note follows how far the pedal is depressed.

Note decays times are pretty short in the looping test, a result at odds with anotherscott's perception of the XS. So we performed another test: I copied the DPBSD looping test to have 4x of them in a row, the first with all notes vel=127, the next 95, then 63 (this is the standard looping test in the DPBSD), then 31. What seems to be happening is this: key velocity is being used as an input variable to the AEG decay rate. This means that decay time is actually quite long when you strike the keys hard, but gets shorter as you hit keys more softly. In real pianos, for a given note, the decay rate is (to a first degree) constant, and decay time varies based on the energy initially input to the string. Since the XS has quite a bit of synth-like flexibility, I belive this behavior can be edited in order to better reflect reality, but it is strange to have this as part of the first factory patch.

anotherscott has this to say about the Motif XS rack piano voice:


We own the ES rack, and I really wish I'd held out for the XS version. The XS front panel makes a lot more sense in terms of layout, and it sports five extra rotary encoders in a matrix formation which must make quick edits much simpler. One thing I wasn't aware of when we ordered the ES is the enormous depth of the housing, which is listed as 372.4mm in the manual. The XS is actually deeper at 379.4mm (probably due to the front panel knobs). Either way, the unit is really deep, and won't fit in most shallow rack rack boxes made for effects.

With the transition from ES to XS Yamaha apparently dropped the dual PLG board expansion capability and top access door, but added an mLAN16E2 (FireWire) option to the rear. Several of the PLG expansion boards look pretty interesting: there's the usual boring smattering of sample expansions, but there's also a DX-like FM synth, an AN1x DSP analog synth, a virtual acoustic synth, and a vocoder - all of which I would love to own, but accessing them via the cryptic ES front panel and menus doesn't seem like it would be very much fun.


Some analysis pix and text review:


Figure 1. Spectral pan view of the pedal sympathetic resonance test, stimulus removed, normalized to -1dB to increase clarity, pedal down @ left, pedal up @ right. Sympathetic resonance is subtle, and sounds kind of fake and echoy to me, particularly with single notes.


Figure 2. Spectral frequency view of the key sympathetic resonance test. No visible or audible key sympathetic resonance.


Figure 3. Waveform view of the extra decay test we performed on the XS, which is the looping test @ vel=127, 95, 63 (the DPBSD standard - highlighted), and 31. Vertical zoom applied to see the noise floor. The decay time is nice and long at higher velocities, and unrealistically short for lower velocities (though this can likely be corrected via editing).


Figure 4. Spectral phase view of the note C3, normalized to -1dB to increase clarity, cursor at attack / loop transition. Attacks are fairly long (as these things go) but loops samples are short and highly processed.


Figure 5. Spectral pan view of the stretch test, middle notes, normalized to -1dB to increase clarity. This voice is fairly stretched, and stretch group transitions are audible over the low and middle range.


Figure 6. Spectral frequency view of the layer test, cursor located at first layer transition. All three layer transitions are audible, with little timber variation within each layer.


Figure 7. Waveform view of the layer test, cursor located at the second layer transition. The third layer transition is also clearly seen here.


------------------------
- Yamaha Motif Rack XS -
------------------------
FILE & SETUP:
- dpbsd_v1.9_yamaha_motif_rack_xs.mp3
- This is the first patch: "Full Concert Grand".
- Setup: Mac with ProTools LE, Mbox interface @ 24-bit, MP3 via Quicktime Pro.
- Recorded by "anotherscott".
PROS:
- Passes the pedal sympathetic resonance test, the effect is subtle.
- Passes the quick partial damping test.
- Passes the half pedaling test.
- Visible velocity layer switches @ vel=50,72,108.
CONS:
- Pedal sympathetic resonance sounds fake: echoy, reverby, buzzy.
- Half pedaling modulates the note decay time but not the pedal sympathetic resonance effect.
- Fails the key sympathetic resonance test.
- Fails the silent replay test (note damps @ pedal up).
- Note decay times are fairly short (~1/2 Pianoteq) @ vel=63, oddly quite a bit longer at higher velocities.
- Attack sample lengths are (C1:C8): 2.7,2.4,2.3,2.2,1.8,1.3,0.8,? seconds.
- Loop sample lengths are (C1:C8): 0.9,0.8,?,?,0.4,0.37,0.45,? seconds.
- Loops sound very static with no "wobble" except for C7 where the loop is quite audible.
- Fairly stretched, visible over the entire range, audible over the low and mid notes.
- Stretch groups: 2,3(x10),2,3,4,3,3,2(x3),4,2,3,1,2,1,2,3(x6),2 = 33 groups.
- Audible velocity layer switches @ vel=50,72,108.
- Velocity layers are largely unblended.
OTHER:
- Notes played @ vel=1 produce no sound.
- Dynamic range 32.5dB (vel=1:127).
- MP3 levels: peak @ -1.4dB, noise floor @ -79dB.
- Date reviewed: 2011-02-19

Is your "silent replay" test the same as my test?

At first I thought it was different, but when I read that the Kawai MP10 failed, it sounded like it failed the same way I've found it to fail (on other DPs, haven't tried the MP10 yet).

If it is the same, I think you should update the results to include whether the sympathetic resonance is cut off when you release the damper while still holding down the silently replayed note. (It should be cut off.)

This was actually the subject of my very first post here! It's so frustrating to find this flaw persisting year after year.

The DPBSD silent replay test is slightly different. In the readme file it is cryptically described like this:

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

Which means:
1. C2 is played at MIDI velocity 100.
2. One second later the damper pedal is fully depressed.
3. One second later C2 is lifted.
4. One second later C2 is played at MIDI velocity 1.
5. One second later the damper pedal is fully lifted.
6. One second later C2 is lifted and G5 is played at MIDI velocity 15 to signal the end of the test.

To fully pass the test, the initial C2 must decay all the way through to the end with no damping. No audible replay at at step 4 is a plus.

OK, I got it.

For the Kawai digitals (primarily concerned with the MP10), does MIDI velocity 1 produce a new note? You might have to crank the volume or zoom in on your waveforms to detect it. And what's the lowest MIDI velocity that produces a new note? I remember trying an EP3, and no matter how slowly I pressed a key, it'd sound a note at the bottom.

jscomposer, the MP10 doesn't, however other Kawai DPs (currently) do.

Cheers,
James
x

Version 1.03 of the RD-700NX firmware has fixed MIDI SYSEX data request messages! I was getting no response before to a data dump request. I updated the NX MIDI post here if anyone is interested. There is a link there to my MIDI test files too.

Roland FP-7F Review



Thanks to Piano World forum member "VivatRudolphus" we have a veritable cornucopia of MP3 test files for the Roland FP-7F - thanks for all the time and effort you put into this VivatRudolphus!

- MP3 -
GP1: http://www.mediafire.com/?pvaus54ndu6n57b
GP2: http://www.mediafire.com/?o71g2xw1ig6t13h
GP3: http://www.mediafire.com/?ltb3bp91zz2raz3
Harpsichord Loop Test: http://www.mediafire.com/?vadzfp97ocotopf
Harpsichord Stretch Test: http://www.mediafire.com/?tu3iarp46vfj9cz
Note Chop Test: http://www.mediafire.com/?542527pl6ubna37
Note Chop Test w/ Backing: http://www.mediafire.com/?vl074n6nnn6kh4t

- PIX -
http://www.mediafire.com/?muujtaidaaqapyy

- MIDI -
http://www.mediafire.com/?o002og310mtnt6o

The main piano voice in the FP-7F sounds to me and tests essentially identically to the the main piano voice in the RD-700NX. Except for pedal sympathetic resonance (which is highly dependent on the effect settings) all of the spectral views are virtually indistinguishable between these two DPs. So if you buy the FP-7F it seems almost certain that you are getting the same quality main piano voice as in the NX. If you want more discussion of that voice itself I'll direct you to the NX review.

All three SN pianos in the FP-7F sound like and appear to be variations of one basic sample set, rather than the three more distinctly different APs in the NX. Though the FP-7F gives you IMO the best voice in the NX, the "Concert Grand" - I'm not sure how useful Studio or Brilliant are in the NX but it is interesting having them in there.

VivatRudolphus has this to say about the FP-7F:

The note chop issue in the FP-7F seems less severe than it was in the NX before the latest v1.03 NX firmware was released. I don't think I can hear it at all unless there is a backing track playing. The layered string drop issue reportedly still exists in the FP-7F though. Let's hope Roland is working on a fix for that.

The harpsichord in the FP-7F is almost certainly NOT SN! It is highly stretched, I believe the loops are very short, and the spectral pan view looks strangely almost like it is mono - though it sounds like stereo and the spectral phase views look normal. Oh well, maybe someday Roland will make a SN harpsichord and stick it in their newest stage piano.

The MIDI files are at the share point as well for those who are interested in turning on all of the features in the FP-7F for recording. You might also want to look here for clues on how to crack this problem in the HP307. This can be a very steep hill to climb, even for someone fairly experienced with MIDI. Roland really should turn all the piano sound effects and resonances on by default for MIDI playback, and have them all editable via the front panel during MIDI playback (like they are in the NX). Users shouldn't have to go out of their way to get the same sound whether playing via the built-in keys or via MIDI (most likely recorded via the built-in keys).

Thanks again to VivatRudolphus, who really went above and beyond in providing us with so many excellent samples of this intriguing new DP!

Some analysis pix and text review:



Figures 1A & 1B. Spectral pan view of the note C2, normalized to -1 dB to bring out detail, RD-700NX "Concert Grand" @ top, FP-7F "Grand Piano1" @ bottom. Almost no visible or audible difference.





Figures 2A thru 2D. Spectral pan view of the stretch test, normalized to -1 dB to bring out detail, mid notes, RD-700NX "Concert Grand" @ top, followed by FP-7F "Grand Piano1", "Grand Piano2", and "Grand Piano3" @ bottom. All have essentially the same spectral signature and sound like they share the same fundamental sample set. Relative to "Grand Piano1", the "Grand Piano2" voice seems to be transposed down one note, and "Grand Piano3" up one note - not a big deal but it's odd, I don't think I've ever encountered this before.


Figure 3. Spectral pan view of the stretch test, normalized to -1 dB to bring out detail, mid notes, FP-7F harpsichord voice. Very highly stretched, this almost certainly isn't SuperNATURAL.


----------------
- Roland FP-7F -
----------------
FILE & SETUP:
- dpbsd_v2.0_roland_fp-7f_gp1.mp3
- SW: MIDI w/ Reaper v3.651, Audio w/ Audacity v1.3.11, WAV=>MP3 w/ LAME v3.98.
- HW: M-Audio MIDISPORT 1x1, Behringer U-PHONO UFO202
- Sequenced and recorded by "VivatRudolphus".
PROS:
- Passes the pedal sympathetic resonance test, the resonance is pleasant sounding.
- Passes the key sympathetic resonance test, strings both lower and higher subtly respond.
- Passes the silent replay test, pedal sympathetic resonance can be heard during it.
- Passes the quick partial damping test.
- Passes the late pedal partial damping test, note decay is caught even after 0.5 seconds.
- Passes the half pedaling test - damping isn't too "buzzy" but sounds real enough.
- Long, fairly natural-sounding note decay (decay times on the order of Pianoteq).
- No visible or audible stretching, notes look random in the wave and phase views.
- No visible or audible layer switches, most timbre change is in the upper 1/2 velocity range.
- Audible duplex scale sympathetic resonance.
- Probably good enough to realistically record solo.
CONS:
- No pedal-down "loom of strings" sound effect via MIDI.
- Duplex scale sympathetic resonance sounds kind of "springy/buzzy".
- No detectable key-up or pedal-up sound effects.
OTHER:
- Notes played @ vel=1 produce no sound.
- Dynamic range 47dB (vel=1:127).
- Dampered | undampered transition: F#6 | G6
- MP3 levels: peak @ -0.6dB, noise floor @ -72dB.
- Date reviewed: 2011-04-02.