Disklavier

Thank you Mick for pointing out this paper.  Although the authors did 
not fully answer their title question, "Are computer-controlled pianos a 
reliable tool in music performance research?", I thought their approach 
and exposition of the timing and sound power errors were excellent. 
What is now needed to complete the answer to their question are 
measurements on human perception:  How much of a timing error or sound 
power error is perceptible?

I am looking forward to one day reading a parallel paper on the Mark III 
and Pro.

--Bob Welcyng

mick hamer wrote:

> Hello all,
> 
> I think that Robert has got the limitations of the Disklavier just about
> right. Some academic researchers published a paper on the accuracy of
> the (Mark II) Disklavier last year. They found that the accuracy of the
> recording was better than the accuracy of the playback. In particular
> the timing accuracy of the playback of soft notes was poor and the
> solenoids did not reproduce high midi velocities (there's no audible
> difference between a midi velocity of 100 and 120). 
>         If anyone's interested in the full details there's a paper
> online with full details of their measurements at http://ftp.ai.univie.a
> c.at/papers/oefai-tr-2001-27.pdf.
>         The original query related to piano roll conversions played back
> on the Disklavier. Piano rolls had their own timing problems and I'd
> have thought that timing problems are much more likely to be a function
> of the original roll (and its conversion to midi) than errors produced
> by the Disklavier.
> Hope this is helpful
> Mick
> 
>>You are quite correct that there are measurable errors in the recording 
>>and reproduction of the Disklavier Mark II.  At least that is what I 
>>have found on mine.  The regulation of my Mark II has been maintained to 
>>the satisfaction of my RPT and the keyboard and pedal calibrations have 
>>been  periodically performed.  The measurable discrepancies between what 
>>is played and what is reproduced are not limited to timing, but also 
>>include loudness.
>>
>>Also, the same file played twice or more consecutively, reproduces 
>>slightly differently each time, both in loudness and timing.
>>
>>The magnitude of the errors (that I have found on mine) is generally 
>>dependent on the note velocity.  For example, with the Volume control 
>>set to "0", notes that were recorded at a velocity of about 100 are 
>>reproduced with a sound power from 1 to 4 dB softer than what was 
>>played; notes that were recorded at a velocity of about 25 are 
>>reproduced at a sound power of from 1 to 5 dB louder than what was 
>>played; notes that were recorded at a velocity of about 60 are 
>>reproduced at a sound power of from 0.5 to 2 dB louder than what was 
>>played.
>>
>>With the Volume control set to "0", the sound output limits for note 
>>velocity values from 93 to 127--that is, there will be no difference in 
>>the loudness of a particular note whether its velocity is 93, 110, or 
>>127.  In order to hear those gradations in dynamics, you would need to 
>>set the Volume control to "-4".
>>
>>Once again, the magnitude of error may or may not be noticeable, 
>>depending upon what is being played and who is listening.  Maximizing 
>>the profit of a product evidently doesn't require satisfying the most 
>>astute consumer.
>>
>>Unfortunately, judging accuracy by ear is a very subjective process. If 
>>you are going to try that, and if you wish to approach the truth, I 
>>suggest a "blind" test wherein the evaluators do not know whether they 
>>are listening to an original performance or to a reproduction and can 
>>only sort them out later through coded names.
>>
>>A comment on playing the old PianoSoft disks on the Pro:  The 
>>information that a Pro recording includes--that is the fine continuous 
>>time record of the key and pedal positions--is absent in an old 
>>PianoSoft disk.  Without the benefit of that information, the Pro's 
>>potential for accurate reproduction cannot be exploited.  To truly 
>>evaluate the Pro, you must listen to a Pro with a Pro recording.  A 
>>PianoSoft file cannot do it justice.
>>
>>Yamaha surely knows its market which is probably why they have focused 
>>on the XG and now the CD features.  However, why, after developing the 
>>Pro and bringing it to market several years ago, Yamaha has not 
>>commissioned solo re-recordings targeted at Pro owners simply confounds me.
>>
>>--Bob Welcyng
>>
>>
>>
>>


-- 
Robert Welcyng
Anchorage, Alaska