Timbre preserve about sms-tools HOT 4 CLOSED

Yes let's do it the same way.

...xavier

Xavier Serra
Music Technology Group
Universitat Pompeu Fabra, Barcelona
http://www.dtic.upf.edu/~xserra/

On 16 Nov 2013, at 22:51, sankalpg [email protected] wrote:

Hi Xavier,

In pitch shift while preserving timbre, the new frequency values outside the old spectrum (range) do not get 0 mag, instead below F0 and above the highest harmonic the spectrum is made constant. This is the way its done in old matlab code. Do we want to do it the same way?

Also there is probably a bug in old matlab code. Instead of Ns in the interp function it should be hNs.

—
Reply to this email directly or view it on GitHub.

from sms-tools.

Sure, done.

Just one more thing, I had to do linear interpolation like the way its done in Matlab codes. Did you add cubic interpolation for some specific purpose? Because for cubic interpolation at every frame at least 4 values are required and in some frames it doesn't find 4 peaks and so it produce glitches.

from sms-tools.

I thought cubic would be better, but yes linear will be fine.

...xavier

Xavier Serra
Music Technology Group
Universitat Pompeu Fabra, Barcelona
http://www.dtic.upf.edu/~xserra/

On 16 Nov 2013, at 22:59, sankalpg [email protected] wrote:

Sure, done.

Just one more thing, I had to do linear interpolation like the way its done in Matlab codes. Did you add cubic interpolation for some specific purpose? Because for parabolic interpolation at every frame at least 4 values are required and in some frames it doesn't find 4 peaks and so it produce glitches.

—
Reply to this email directly or view it on GitHub.

from sms-tools.

closing.

from sms-tools.