[sdiy] MIDI volume to volts formula

[rsdio at audiobanshee.com]

On Jun 6, 2018, at 6:07 PM, Adam Inglis <21pointy at tpg.com.au> wrote:
> On 7 Jun 2018, at 4:48 AM, Ben Bradley <ben.pi.bradley at gmail.com> wrote:
>> On the other hand, I think a piano is a roughly linear function of
>> input velocity (or force) and "signal" output. This may require
>> "original research."
> 
> There must be a stack of research out there, possibly some of it in-house/proprietary by the big instrument makers, regarding this.
> I would’ve thought you be better served by a sinusoidal function, rather than log or linear, but of course it is going to depend mostly on the qualities of the particular tone or patch you are controlling at any one time (even more than personal playing style).

Why sinusoidal? Are you suggesting the volume should eventually decrease and then go negative as velocity increases?

I suppose you might be talking about the curve formed by the first quadrant of a sine wave, as opposed to the full periodic cycle.

In any case, you’re surely correct about the research. I recall that Yamaha or some other digital piano company built a robot that played every key of an 88 grand at each of 127 velocities, separately recording each of the resulting sounds in high quality digital audio for analysis. Somehow, they extracted sufficient modeling parameters from those tens of thousands of recordings so that resynthesis was possible rather than brute-force sample storage and playback. The latter was the purview of the GigaSampler product.

Brian Willoughby
Sound Consulting