Forums > MaxMSP

vocoding strings

January 13, 2009 | 12:06 am

Hi all,

Does anyone know of an MSP patch that can transpose audio generated from a single guitar string? I know it’s tricky — simple frequency-domain methods cause phase smearing, but in the time domain one has to identify transients, treat different bandwidths differently, identify the pitch in order to determine the size of granules, etc. I know enough to be pretty confident it’s beyond me, but I’d love to find one if someone else has made it already …

My VG-99 guitar processor can do this, for each string independently, although it has the unfortunate property that changing pitch changes the apparent tension on the string — e.g. pitching it down makes it sound all detuned, etc. (Also, they don’t say how they did it, which is arguably much more annoying.)

I’ve found a lot of stuff on the net regarding phase vocoding for voice. I feel like the problem for strings ought to be easier, because string timbres are more harmonic (that is, the overtones hardly deviate at all from integer multiples of the fundamental) and because transients are more clearly identifiable. But I haven’t been able to find much written about the corresponding problem for strings.

Best,
Jeff


January 13, 2009 | 12:41 am

[fiddle~] is a really fun object. I think it does what you want it
to, probably.

http://crca.ucsd.edu/~tapel/software.html

On Mon, Jan 12, 2009 at 4:06 PM, Jeff Brown wrote:
>
> Hi all,
>
> Does anyone know of an MSP patch that can transpose audio generated from a single guitar string? I know it’s tricky — simple frequency-domain methods cause phase smearing, but in the time domain one has to identify transients, treat different bandwidths differently, identify the pitch in order to determine the size of granules, etc. I know enough to be pretty confident it’s beyond me, but I’d love to find one if someone else has made it already …
>
> My VG-99 guitar processor can do this, for each string independently, although it has the unfortunate property that changing pitch changes the apparent tension on the string — e.g. pitching it down makes it sound all detuned, etc. (Also, they don’t say how they did it, which is arguably much more annoying.)
>
> I’ve found a lot of stuff on the net regarding phase vocoding for voice. I feel like the problem for strings ought to be easier, because string timbres are more harmonic (that is, the overtones hardly deviate at all from integer multiples of the fundamental) and because transients are more clearly identifiable. But I haven’t been able to find much written about the corresponding problem for strings.
>
> Best,
> Jeff
>
>


January 13, 2009 | 3:20 am


January 13, 2009 | 4:36 am


Viewing 4 posts - 1 through 4 (of 4 total)