Disklavier

I have an older Disklavier, model HQ100, which I drive from a PC 
using WinAmp via a USB MIDI adaptor and the MIDI input port.  The 
only volume control is the knob that adjusts the distance of the 
hammers from the strings, but that has no effect on the force with 
which the keys are struck, which depends solely on the MIDI velocity 
value.

I would like to have a program to adjust MIDI velocity levels "on the 
fly", and ideally one that would work in conjuction with WinAmp (but 
that is not a requirement as long as there would be some means to 
queue files in a playlist).  I sat at dinner last night thinking of 
the velocity adjustment algorithm to use.  On arriving home, I 
decided to check this user group to see if anyone had already solved 
this problem, and I found your timely message thread.  It seems the 
problem is not yet solved, but there are others out there who have 
similar requirements.

I can contribute the velocity adjustment algorithm, if someone else 
who understands MIDI file structure can do the programming.  I have 
written a small Excel spreadsheet with interactive slider bar inputs 
and a velocity distribution graph so that you can see the impact of 
the variables on the velocity value modifications.  (I don't see how 
I can attach the file here, but I will try to post it in the "all new 
uploads go here" folder.  If anyone wants it, write to me outside the 
group forum and I will send it.)

Briefly, there are three variables:  

1) Absolute Maximum Velocity is the value that under no circumstances 
can be exceeded.  This is the safety value so that high velocity 
values do not cause damage to the piano.  Maximum MIDI velocity is 
127, however values of 100 to 110 have been suggested as 
maximum "safe" levels.  MIDI velocities above this maximum will be 
truncated to this value.

2) Target Minimum Velocity is the lower threshhold for sounding the 
piano, or for setting a minimum volume higher if so desired.  All 
MIDI velocities below this level will be raised to this level, unless 
volume control is set lower.

3)  Volume Control is a percentage of original velocity, within the 
bounds set by Absolute Maximum Velocity and Target Minimum Velocity.  
If Volume may be adjusted below the Target Minimum Velocity if 
desired to turn volume down completely.  Volume profile will be 
completely flat at this point.

The algorithm (in Excel format) is: 
=MIN(AbsMax,MAX(TargetMin,(TargetMax-TargetMin)/(OrigMax-TargetMin)*
(OrigVel-TargetMin)+TargetMin))

Where: 
AbsMax = Absolute Maximum Velocity (safety limit)
TargetMin = Target Minimum Velocity (lower threshhold)
TargetMax = Volume Control % times 127 (maximum MIDI velocity value)
OrigMax = 127 (maximum MIDI velocity value)
OrigVel = Input velocity value from MIDI file (the value to be 
modified).

This will make more sense if you view the file, play with the 
variable slider bars, and see from the graph how the algorithm works.

Future enhancements could include a curved profile (this one is 
linear), different profiles or settings for various ranges (treble, 
mid-level, bass), etc, but I think the algorithm I have proposed will 
be more than satisfactory for 99.44% of users.

Is there anyone in the group that could program this?    

John Sutton
China


--- In disklavier@yahoogroups.com, Carl Youngblood <carl@y...> wrote:
>
> It seems like the algorithm being used is wrong.  If a note is 
already
> soft, it should not be reduced to zero.  It should be reduced less 
the
> further it is from the desired overall volume, until turning down 
the
> volume doesn't change it at all.  Think of a normal distribution, 
with
> the desired volume being right at the middle of the bell curve. 
> Turning down the volume should bring notes closer to the center of 
the
> bell curve instead of just subtracting uniformly from their
> velocities.
> 
> On 12/10/05, bnbwv <davis2002@c...> wrote:
> > In trying to reduce the volume on my old wagon grand, there are 
two
> > complications.  First, the volume controller (7) has no effect on
> > hammer velocity (volume).  Second, if you just scale down all
> > velocities, the softer notes won't sound at all.  You need to map
> > velocities into a range from a low of around 25 to a high that
> > depends on the 'volume' you desire.  Your MIDIMOD program is what 
is
> > needed, with the mapping being done on the fly rather than as a 
file
> > in / file out operation.
>