Csound Csound-dev Csound-tekno Search About

[Csnd] pvscale glissando problem

Date2010-12-03 18:29
Fromjoachim heintz
Subject[Csnd] pvscale glissando problem
when i use the pvscale opcode with a continuous rising or falling  
kscale value, i get kind of jumps. i suppose these may result in  
transitions between the bins.
i have compared this with the maxmsp gizmo~ object which gives very  
good results.
this is the maxmsp result (glissando from one octave higher to one  
octave lower than the original pitch of the sample):
http://www.joachimheintz.de/unsicher/gliss_maxmsp.wav
this is the pvscale result with the same arguments as the gizmo test  
(fftsize=4096, overlap=4):
http://www.joachimheintz.de/unsicher/gliss_csound_4096-1024.wav
if i use the usual values (fftsize=1024, overlap=4), the transitions  
are more noticeable:
http://www.joachimheintz.de/unsicher/gliss_csound_1024-256.wav
the values fftsize=1024, overlap=16 give better results, but are still  
worse than the gizmo output:
http://www.joachimheintz.de/unsicher/gliss_csound_4096-256.wav

the gizmo~ object is written by richard dudas who decribed the method  
in the icmc proceedings 2002 as "floating formant" method:
http://quod.lib.umich.edu/cgi/p/pod/dod-idx?c=icmc;idno=bbp2372.2002.028

would it be possible to implement this method as an option in pvscale?  
or is there another pvs opcode which has good results for such a  
glissando situation?

thanks -

	joachim


ps this is the sample:
http://www.joachimheintz.de/unsicher/OrgelMono.wav

this is my csd:


sr = 44100
ksmps = 32
nchnls = 1
0dbfs = 1
instr 1
ain soundin "OrgelMono.wav"
ktrans expon 2, p3, .5
fsig pvsanal ain, 1024, 256, 1024, 1
ftrans pvscale fsig, ktrans
aout pvsynth ftrans
out aout
endin


i 1 0 10




Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

Date2010-12-03 19:05
FromVictor Lazzarini
Subject[Csnd] Re: pvscale glissando problem
well, there are some improvements to pvscale regarding formants, a new  
formant warping opcode and an experimental
locking opcode. Maybe with a combination of these something similar  
can be concoted. I will also need to look at the paper
in a bit more detail to see if it gives enough information for coding.
There are also other opcodes for table reading etc that transpose  
pitch differently and might yield a better result.


Victor
On 3 Dec 2010, at 18:29, joachim heintz wrote:

> when i use the pvscale opcode with a continuous rising or falling  
> kscale value, i get kind of jumps. i suppose these may result in  
> transitions between the bins.
> i have compared this with the maxmsp gizmo~ object which gives very  
> good results.
> this is the maxmsp result (glissando from one octave higher to one  
> octave lower than the original pitch of the sample):
> http://www.joachimheintz.de/unsicher/gliss_maxmsp.wav
> this is the pvscale result with the same arguments as the gizmo test  
> (fftsize=4096, overlap=4):
> http://www.joachimheintz.de/unsicher/gliss_csound_4096-1024.wav
> if i use the usual values (fftsize=1024, overlap=4), the transitions  
> are more noticeable:
> http://www.joachimheintz.de/unsicher/gliss_csound_1024-256.wav
> the values fftsize=1024, overlap=16 give better results, but are  
> still worse than the gizmo output:
> http://www.joachimheintz.de/unsicher/gliss_csound_4096-256.wav
>
> the gizmo~ object is written by richard dudas who decribed the  
> method in the icmc proceedings 2002 as "floating formant" method:
> http://quod.lib.umich.edu/cgi/p/pod/dod-idx?c=icmc;idno=bbp2372.2002.028
>
> would it be possible to implement this method as an option in  
> pvscale? or is there another pvs opcode which has good results for  
> such a glissando situation?
>
> thanks -
>
> 	joachim
>
>
> ps this is the sample:
> http://www.joachimheintz.de/unsicher/OrgelMono.wav
>
> this is my csd:
> 
> 
> sr = 44100
> ksmps = 32
> nchnls = 1
> 0dbfs = 1
> instr 1
> ain soundin "OrgelMono.wav"
> ktrans expon 2, p3, .5
> fsig pvsanal ain, 1024, 256, 1024, 1
> ftrans pvscale fsig, ktrans
> aout pvsynth ftrans
> out aout
> endin
> 
> 
> i 1 0 10
> 
> 
>
>
> Send bugs reports to the Sourceforge bug tracker
>           https://sourceforge.net/tracker/?group_id=81968&atid=564599
> Discussions of bugs and features can be posted here
> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
> "unsubscribe csound"
>



Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"