Re: OT: StringPort for polyphonic guitars

[andy butler <akbutler@tiscali.co.uk>]

Stefan Tiedje wrote:
> andy butler schrieb:
>> so the latency is greater than a hardware Midi converter
>> which doesn't have to buffer the audio in and out of a pc/mac.
> 
> The audio buffer isn't the main latency source if you do pitch to Midi, 
> its the algorithm.

indeed, my point is that the buffer latency is additional
to any pitch calculation 

> The neural networks used in hardware controllers can 
> track the pitch within a quarter wave length, if they assume that it is 
> a guitar or bass string.

Thinking about the physics,
the limit is actually the speed with which the impulse of the pick hitting
the string travels up to the fret and all the way back down to the p/u.
The speed of that pulse is such that it travels twice the string length
in a single wavelength, so to carry any information about which fret
is used to the p/u takes more like 7/8 of a wavelength.

Unless of course the guys from Axon know different ;-)
...but note that the Axon device can also derive a controller 
value from the position of the pick along the string.

afaik, the Roland device doesn't have the Axon algorithm, and is 
a bit slower.


> Other methods need several complete cycles to 
> determine the pitch. Higher latency for lower notes...

as they only need to guess to the nearest semitone, they
can manage with a single cycle (or nearly).
Generally they send midi-pitch bend as well, to correct a wrong guess.


> 
> But as I see it the main point of the string port would be to get the 
> audio signal straight into the computer and you decide what to do with 
> it. I would send each string to a different delay line/looper for 
> example...

:-)

and this could somehow be explained to manufacturers one day


> Stefan
> 
andy