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

Yes, I agree, to ensure continuity when hearing, they need to flatten the variation of 1)amplitudes and 2)directions of harmonics. It should be quite difficult to make a "seamless pasting", considering the 2 dimensions of the problem... Probably a painful random bricolage, ending with a massively processed sound more easy to paste...

This picture is for me the confirmation that the "physical presence of sound", is still different between AP and DP. Spatialization of sound (both during recording, and playing) is really important, to capture liveliness and depth of a tone (the harmonics "swing" and roll in the box). Other point is filtering low and high frequencies is also limiting the perception of vibrations of the piano by our body...

dewster,

Do you have some pan view of a note played on an accoustic piano ? I am very interested to compare, with DP samples. This stereo panning is really intriguing for me.

Do you know where it comes from, it is a slow directional sound motion with back and force, nothing that can come from a reverb, so seems to be a mobile sound source : sound board ?

For me this is the vibration of the sound board inited at oscillting string fix point, and making wave in the sound board. Correct ?

I previously did a comparison between Roland SN, Kawai UPHI, and a fully sampled Yamaha C7 from VintAudio here.


I believe the slow, rather random pan is mostly from having three unison strings tuned slightly differently beating together in close proximity to each other on the soundboard. Some of it could also be the primary modes of excitation rotating around the axis of the string. I haven't extensively researched what causes this but it is certainly there on a real piano and short loops understandably kill it every time.

Thanks, perfect

Dewster, I am lost...
Please try again (differently ), I really want to understand that...

Perhaps slow traveling mode (emerging from quick stationary modes) :
http://www.youtube.com/watch?v=GMXlzCX5gio

Or what about stationary vibrating modes drifting or being instable (between 2 modes) :
http://www.youtube.com/watch?v=GtiSCBXbHAg&feature=related

Look a this increadible cimatics example with human voice :
http://www.youtube.com/watch?v=KU84ckD1AcA&feature=related

When you play a note on a piano, except for the lowest bass notes, the hammer hits more than one string. Most notes in fact have three strings. These strings are tuned almost exactly the same, but the slight difference produces a slow moderate phasing sound.

Another thing that might cause this phasing is the direction the strings are vibrating. On a grand piano the hammer hits the strings vertically, but during the decay I think the vibrations can slowly rotate around to horizontal and back again to vertical.

Thanks

dewster,

Phasing sound is a "new" sound, with its own lower frequency.

In your pan view, all the curves are moving. If it were phasing sound, as 1 note is played (max 2 strings vibrating), at most one or 2 phasing sounds could appear and slide. Or, all harmonics will generate there own phasing sound. But even in that case, some curves should remain static (that is at least 2 statics harmonics for 1 phasing sound).

Other thing that is strange, the moving frequencies change of direction (inflexion) in random area. I am not sure a phasing sound can do that.

Actually eitheir moving wave reverb (or not) against the edge of the soundboard, or at against soundboard barriers. Probably barriers are like guide for "almost"stationary modes (like guide for chaldni nodes in guitar sounboard). Or this "quasi" nodes moves really "randomly"...

I don't know, I don't understand...

When you add together three sine waves with slightly different frequencies, the result can be a somewhat random looking wave. This is essentially what's going on with the piano notes: their stereo harmonics sum together to make new harmonics with a somewhat random stereo image. This harmonic sum appears as a single line in the spectral pan view, drifting up and down in the stereo field. I'm sure some coupling is going on through the soundboard as well. It's all rather complex in real life, which is why it's probably best just to sample this stuff than to simulate it.

I think I understand a little bit more.

I imagine the 3 strings vibrating at almost the same frequencies, and almost the same phase, coming for almost the same place.

All these sources (without considering coupling for the moment), generates waves in the sound board (either starting from a point or a segment, depending of the rigidity of the chevalet (where string are fixed), so circular waves from spot, or transerval waves from segment.

These is true for all 3 fundamental waves, and all 3 waves per harmonics transmitted by string (not considering sound board is an "active" transmitter, and may modify waves (take more energy for some, re-create new upper harmonics, etc).

All these waves are moving 5000m/s in sound board, rebounding (reverbing) against edge, then as edge are not regular, all these echoes supperpose together.

We have then complex interference pattern beating, increasing amplitude in some place, decreasing in another place, and probably, these patterns (with high up front), still beating at harmonic frequency, due to slight phasing effects, can begin to drift a little (1m per second or so) in random direction in the sound board, in the beating mesh, creating stereophonic effect...

Is that description correct ?

PS : I want to understand how complex is the modeling of sound board...

Regards

And sometimes, for certain frequencies, all interference pattern (or at leat for fondamental) stabilize/synchronize together, and we get a chaldny stationary mode ?

Still correct ?

I'm looking seriously at the Kawai CS9 which also has UPHI - is this going to be exactly the same results in terms of Attack sample length as the CA63 test you have done? Slightly dissapointing if so as the Avant Grand is a few seconds better for most of them Please let us know.

Kind Regards,
Allan

Good question. If you can get near one long enough to render the DPBSD MIDI to MP3 to I'd be able to tell you. From the manual:

http://www.kawai.co.jp/worldwide/support/manuals/pdf/CS9_EN_20110613.pdf

Page 45 describes how to convert a MIDI file to an MP3. All you need is a thumb drive with the DPBSD MIDI file on it. Kawais generally have quite nice stand-alone rendering abilities, kudos to them for that.


The DPBSD testing for me has been one long trail of tears with only a few bright spots here and there. The products seem to be improving, but so slowly only a turtle or a great sequoia might notice. We haven't quite reached the "big shakeup" in the DP market (when unlooped & unstretched high-end models will be the norm). As with all computing equipment, if you can afford to wait you can likely get more for less later.

2 dewster
Just have seen your begging section.
Well, I have LIMEX Vienna Grand Piano expansion board. And I'll try to post here the result mp3. I am sure the expansion board has the same chipset as the piano module.
Just two words: ATSAM9708 + 256MB + 3xPCM1754 + AL3201B. No magic.

SOMEWHERE there are some slow mo movies of this.
The closest description I can offer is that it is somewhat similar to the motion that I have seen in a skip rope when a child at one end swings it in a circle.
Not really a "traveling screw" wave motion, but certainly more than merely vertical.
As I recall the explanation given was that the initial vertical motion "collides" with sound energy from the soundboard, which deflects the string vibrations, resulting in some horizontal component.

This was a long time ago, the details escape me, it may still be found with a well crafted search (-:

Hokey-smokes, I had no idea anything like the ATSAM9708 (dual DSP / GPU to support 128 polyphony + effects) even existed outside of the proprietary realm, thanks for that! The AL3201B (digital reverb engine) is also a very interesting part.

I'll gladly take your MP3 whenever you might be able to provide it.

The thing that really put me in a confusion is a 2$ DAC onboard. It spoils all the impression here, don't you think so? When it comes to a compact layout, AD1854 of my RP-X takes a slightly more space.

The PCM1754 looks nice enough on paper. Certainly DACs these days are lightyears ahead of what we had just a decade or so ago, and with DACs showing up in literally everything price may not necessarily be a good indication of performance. Beyond hearing some aliasing when trying to reproduce high frequency, high amplitude sine waves with my old PC sound card, I can't say I've ever directly experienced a "bad" DAC. Perhaps I'm not able to hear it for some reason, or the opportunity just hasn't presented itself?

It's been said that the very worst DAC is better than the very best speaker.

Greg.

Absolutely. Though it's kind of amazing speakers have been refined to the point they are currently. Who'd think a piece of paper set in motion by an electromagnet and placed in a box could sound so good?

The most likely limiter of magic on the LIMEX Vienna Grand Piano expansion board magic is this: 256MB