have you ever taken about consumer electronics? As an example, Sony might make a line of five DVD players that all look alike except for the higher end, more expensive models might have some extra buttons to allow access to more features.

Take them apart and you find the circuit boards are identical and the low cost model has a blank place where a switch was not soldered down. I'd not be surprised if Yamaha used the same electronics in several DPs.

I think "Stereo sustain sample" is maybe what I call a "pedal-down" sample with a "loom of strings" sound - when the damper pedal is pressed in a real piano it excites all of the strings slightly.

What they call "key-off" I call "key-up" and I think that's clear. Sometimes it's a knock sound, sometimes it's the buzzy sound of a string being damped. Same for the "pedal-up sound".

"Damper resonance" strikes me as insufficiently descriptive and also confusing. Though I can see why they are trying to differentiate what is usually implemented as a global delay effect from something that is more complex - what they term "string resonance".

You do realize how dangerous these statements are above, don't you?

Soon, dewster will not be requesting us to send him mp3 files but to open up our cases and photograph our circuit boards!

The third generation of the thread will have us all jockeying with soldering irons...

Not to mention that colleen/snazzy (the same person) will then go metaphysical-on-our-a** and will further demean dewster for violating the "soul" of the digital instrument by "opening it up" (sic) to even further scrutiny! They prefer darkness to light, and sit in wonderment at the marketing departments across the globe.

If DPs have souls, does that mean we'll have them in heaven?

If my DP breaks, do I need to give it a proper church service and burial?

Yamaha CLP-330 Visual Results

Below are some of the results of Yamaha CLP-330. The text review was posted earlier in this thread.

Kind of disappointing for a Clavinova class instrument.


Spectral pan view of the note C4. Cursor at end of sample / start of loop. Sample length is 1.9 seconds, loop length is 0.5 seconds. Loop periods here are not easy to see and require compression to reveal the details. I decided to not do that here so that the sample / loop transition would be clearer.


Spectral pan view of the note C5. Cursor at end of sample / start of loop. Sample length is 0.8 seconds, loop length is 0.8 seconds.


Spectral pan view of the note C7. Cursor at end of sample / start of loop. Sample length is 0.8 seconds, loop length is 0.2 seconds.


Spectral phase view the layer test. Smooth timbral variation with no visible or audible layer switching (to me; however the owner reports bad and audible layer switching).


Spectral pan view the stretch test. The mid notes are shown here. Stretching is present and audible over the entire range, which is unusual.

These and more pix for the CLP-330, as well as the text review and MP3, are posted at the sharepoint for those who are interested.

If anyone can send me a DPBSD MP3 of the CLP-370 or 380 that would be great.

I have a theory about Clavinovas, and wonder if others think the same way.

These are "cabinetry" DPs, so people are most likely buying them to put in living rooms and such because they feel stage pianos or in-between units like the P-155 are too ugly. Nothing wrong with that, as long as people realize they're paying a hefty premium for it.

Anyway, "cabinetry" DPs usually have fairly decent amp/speaker sound systems in them (as DPs go, that is - in actuality they are rather far from ideal) that people like because they want to play without a lot of fuss and bother. Playing through flat headphones, or particularly through external flat active studio monitors, would be unusual. I think also the "cabinetry" DP buyer tends to be coming at a DP purchase from more of an acoustic piano angle, rather than a sampler angle, and therefore tends to be less savvy in terms of the technical limitations of DP sounds.

Manufacturers know this, and they also know the rather lame built-in sound system will mask many otherwise audible issues with the sample playback. So they feel a bit freer to hack away on the sample set, using shorter and fewer attack and loop samples, cutting corners on the sympathetic resonance and reverb algorithms, etc.

Does that sound crazy? It certainly seems to be the case for the CLP-330 anyway.

yes it does. Just contact the manufacturer directly, you should get some kind of answer back, and if you're lucky you'll get an answer from one of the techs. My feeling is that the current set of pianos is it is all about hybrid technology, half samples, half modeling. And because hybrid tech is in its infancy, piano makers tend to use fewer samples than they should, because they want to show off their modeling software.

GEM did the hybrid thing YEARS ago, as in almost 10 years ago. Few people cared though. I don't know why. They went outta business. So now Yammy is jumping on the bandwagon.

Drewster, thanks for redoing Pianoteq. Just wanna point out that key release is present while pedaling. Maybe it's more difficult to hear with all the resonance, but if you increase the key release noise parameter, you'll hear it's there.

I think GEM's claim to fame was that they were way ahead of everyone else in the sympathetic resonance department. In their earlier models they used a single layer sample and filtering or something, which was interesting too. But as much as I wanted to get on the GEM bandwagon, their samples were still sounded like they were looped with short decay, which killed it for me and perhaps others.

I've reached the point where I really can't justify spending any money on weak or fake sound of any sort, particularly when $500 of PC hardware and $300 of software will routinely blow the doors off almost all DPs out there, some of which cost over five times as much.

I think you have the target market for this kind of piano correct. Most buyers will not be experts in the technology but that applies to buyers of TV sets, computers and even car tires.

But I don't thing Yamaha trys to mask the sound with a poor audio section. On the contrary. They use well designed sound systems because this is what matters. People judge the piano sound based on loudness and the amount of bass.

As an example most people will rate the YDP223 sound as better then the YDP160 even if the 160 has better specs simply because the 223 has better amps and speakers.

Looking at the audio sections of Yamaha DPs they seem to be very good designs in that they get quite good sound from very low cost parts. This is how engineers earn their pay. Any idiot can get high quality with a high budget.

A trick used by commissioned stereo salesmen is to turn the more expensive system up louder. Buyers always think "louder is better". You need to be very careful to match volume levels when you A/B test audio systems so as not to get suckered by that effect.

We only see a minority of buyers here, the owes who think to research a DP. Most I'm sure buy based on price and physical appearance. And necessary low price either. People generally are looking for a match to their budget.

I'm talking more about the treble, I should have been clearer. And I don't think they set out to make a poor audio section, the treble is often rather neglected, particularly when the speakers are not directly facing the listener. But I believe they could easily capitalize on that fact by cutting corners in the sample department until compression artifacting starts becoming audible.


You must be blessed with a higher quality selection of idiots than those that surround me!


Exactly. I've read that even 0.1dB change in loudness can influence how the sound of something is perceived in terms of quality.


Very good point.

Hmmmm... It doesn't look like the document has been released yet.

I don't know if you saw this (or that it will help while you are waiting for the HP-307 manual), but there is a MIDI doc for the HP-203/205, where they list the various configuration settings for "Damper Resonance": their description seems to match your definition of "sympathetic resonance".



Here is the document:

http://media.rolandus.com/manuals/HP-203_MI.pdf

Perhaps some of the Damper Resonance settings for the HP-203/205 would work on the HP-307?

[It seems strange that this document doesn't apply to the HP-207. Maybe Roland has different MIDI EFX implementations for different DPs, even within the same family]

I sent an email to Roland Poland (where I live) asking about the implementation of damper resonance and noise effect. I didn't ask for midi implementation sheet but just why it doesnt work when playing from midi and if it's intentional or can be fixed somehow. If I receive an answer I'll post here.

Hahaaa! Roland Poland! That's like Kawai Hawaii...

Sorry, couldn't resist!

...if only KAWAI had an office in Hawaii!

James
x

I'm from Poland there is also Roland Holland (Netherlands). Hehe!

...Yamaha Panama...

From first answer they seem to be unaware of the problem and ask me if I changed any settings... I asked them to do a specific sequence of operations to try to reproduce it.

I hope the email I got from them wasn't just a brush-off.


Thanks for the pointer! Hard to say. I wish I had one here to perform experiments on.

Thanks! Maybe if enough people ask they will release the secret SYSEX codes - or a firmware update?

To me this is a rather large issue. Having a DP voice that is possibly good enough to record with, but not having certain features of that voice play back via MIDI kind of defeats the purpose. It would be great to be able to have someone perform on it (with good headphones for accurate, balanced feedback to the player), record the MIDI, do a few minor edits, then play it back and record the audio.

DPBSD v1.5 Released

There is now a new version of the DPBSD MIDI file, please use it for all current and new testing. The readme file has been updated as well to explain the new tests, and it contains some extra info on recording details as well.

Basically there are two new tests: one for limited sympathetic resonance of silently held notes, the other for pedal down silent replay behavior.

Some of the tests have been moved to better group like tests, and to place the potentially loudest test first.

====================
= Revision History =
====================
v1.5 - 2010-03-06:
- Moved pedal down sympathetic resonance test to be first test due to high amplitudes involved.
- Changed pedal down sympathetic resonance test timing & notes - now more dissonant & excludes damperless keys.
- Third phase of pedal down sympathetic resonance test now has keys down for full time to match second phase.
- Third phase of pedal down sympathetic resonance test pedal up event moved a bit later than key-up events.
- Added key down sympathetic resonance test.
- Added pedal down silent replay test.
- Moved partial pedal test to 4th to better group these similar tests.
- Shortened C8 by 1/3, C9 by 1/2 in looping test to better match Pianoteq decay time.
- Reduced key down time from 120 to 100 for all notes in velocity layer test to remove event ambiguity.
- Can use recording level (test 0) to reveal unusual note repeats (e.g. odd damped behavior of Yamaha CLP-330)

v1.4 - 2010-02-17:
- Added C5 (x7) velocity = 127 at start of file as an aid in setting peak recording level.
- Added two pedal=111 notes to the partial pedaling test, moved all pedal events midway between the notes.

v1.3 - 2010-02-03:
- Fixed problem with a loud velocity in the middle of test #3 (was double note).
- Extended sustain (key down) times for second phase of test #4 - now the same as third phase.

v1.2 - 2010-02-03:
- Changed velocity from 63 to 100 for all notes in test #4.
- Created readme file.
- There seems to be a problem with a loud velocity in the middle of test #3?

v1.1 - 2010-02-02:
- Born.

I found the link to the sustain test (aka pedal down silent replay test), thanks theJourney (and kudos to jscomposer):

http://www.pianoworld.com/forum/ubbthreads.php/topics/703018/1.html

I saw your post over on the other thread as well.

Thanks!

I think most people who would record to MIDI and edit would prefer to play back the MIDI through a software instrument. Not only are these better samples but the technical quality is very good because there is "analog gap" the virtual instrument goes straight to a .wav file.

I agree, but then you lose the tracking between how someone plays on a particular DP and the playback on something entirely different. I've had to deal with that before and it's a real pain. Even different speakers or headphones while the person is playing can drastically alter the sound of the DP (particularly the bass), and therefore how they play it. Record MIDI => Playback on same device is the way to go.

Yamaha AvantGrand N3 Review

Volusiano was kind enough to run his Yamaha AvantGrand N3 through the DPBSD gauntlet of tests. I received the MP3 file yesterday and spent some time analyzing it last night and this morning.

The samples are the longest I've seen so far in something that is conventionally looped, and the decay time is nice and long. It would really benefit from longer loop samples though. It appears to be fully sampled (no stretching), and the velocity timbre transition is smooth. Pedal down sympathetic resonance sounds better than what Yamaha usually does.

------------------------
- Yamaha AvantGrand N3 -
------------------------
FILE & SETUP:
- dp_bsd_v1.5_yamaha_avant_grand_n3.mp3
- USB Flash drive (MIDI file), headphone out, Laptop, Adobe Audition.
PROS:
- Passes the pedal down sympathetic resonance test - resonance is fairly nice.
- Pedal down sample is a realistic "loom of strings" sound.
- I believe it passes the key down sympathetic resonance test, but the effect is very subtle.
- Passes the pedal down silent replay test.
- Responds to partial pedaling.
- Highly damped partially pedaled notes have buzzy partially damped sound.
- Key-up of louder notes produces realistic buzzy damping sound.
- Sample lengths are somewhat longer than usual.
- Nice long decays on the order of Pianoteq.
- Sample lengths are (C2:C9): 4.4,4.3,3.9,3.5,3.0,2.1,?,? seconds.
- No visible or audible stretching.
- Velocity appears to be a smoothly blended multi-layer sample set.
- No audible velocity layer switching.
- Something like a velocity layer switch barely visible @ vel=78 (spectral pan view).
- Huge dynamic range (~55dB, vel=1:127).
CONS:
- Obviously looped, both visually and audibly, particularly the lowest notes, though it isn't too badly done.
- Loop lengths are rather short.
- Loop lengths are (C2:C9): 0.7,0.67,0.67,0.67,0.56,0.43,?,? seconds.
OTHER:
- MP3 levels: peak @ -1dB, noise floor @ -70dB.
- Date reviewed: 2010-03-08



Spectral pan view of note C4. Loop begins at cursor.


Spectral pan view of note C4. Single loop is highlighted.


Spectral phase view of note C5. Loop begins at cursor.


Spectral pan view of note C5. Single loop is highlighted.


Spectral phase view of stretching test, the bass notes are shown here. No evidence of stretching anywhere on the keyboard.


Spectral frequency view of the layer test. Velocity layers are smoothly blended with no visual or audible steps.


Spectral phase view of the layer test.

As usual, there are more pix, the MP3, and the test review located at the MediaFire sharepoint.

Interesting stuff!

Thanks dewster and Volusiano.

Cheers,
James
x

Thanks! It was quite interesting reviewing such a high-end DP. It was also the maiden voyage of v1.5 of the test.

Yey - was hoping for the N3 - thanks to Volusiano (and dewster)!

It is interesting that the N3 does better, technically, than the CP1/CP5, in this test. But still, there's room for further improvements...

Now the one sample I'm still waiting for is Roland HP207 or RD700GX.