how can i make make this little patch more cpu friendly. i want to use it multiple times in a big patch i'm building.

gizmo-cpu-usage-question

ya, i made the comparison within the pitchshifting-playground patch which jean-francois mentioned and chose the transposer object instead of gizmo~ for my own needs because the doppler technique it uses seems to use half the CPU of gizmo's FFT technique. Personally, if I'm pitch-shifting, I'm not really concerned with artifacts appearing while making radical shifts so gizmo~ is a bit of overkill for me.

Max5(or4.6)/examples/effects/transposer-folder/lib/transposer.maxpat

check it out:
Max5(or4.6)/examples/effects/transposer-folder/lib/transposer.maxpat


Among the other things that you might try (appart from changing the 

window size) is to add some threaholding mechanism before sending the 

real/imag to gizmo~. In order to transpose there is a peak analysis 

which is computed and if the ultra low amplitude are not present it 

will help. You might not gain much if you do only one transposition 

but with multiple gizmo~ in a pfft~ it's definitely worth to try.

On 27 nov. 08, at 09:43, Pierre Alexandre Tremblay 

> Don't forget the pfft~ 4th argument if you use many, it helps to 

> share the load by de-syncing the fft analysis

>> ya, i made the comparison within the pitchshifting-playground patch 

>> which jean-francois mentioned and chose the transposer object 

>> instead of gizmo~ for my own needs because the doppler technique it 

>> uses seems to use half the CPU of gizmo's FFT technique. 

>> Personally, if I'm pitch-shifting, I'm not really concerned with 

>> artifacts appearing while making radical shifts so gizmo~ is a bit 

>> Max5(or4.6)/examples/effects/transposer-folder/lib/transposer.maxpat

Among the other things that you might try (appart from changing the  
window size) is to add some threaholding mechanism before sending the  
real/imag to gizmo~. In order to transpose there is a peak analysis  
which is computed and if the ultra low amplitude are not present it  
will help. You might not gain much if you do only one transposition  
but with multiple gizmo~ in a pfft~ it's definitely worth to try.

On 27 nov. 08, at 09:43, Pierre Alexandre Tremblay  
 wrote:

> Don't forget the pfft~ 4th argument if you use many, it helps to  
> share the load by de-syncing the fft analysis
>
> pa
>
> Le 08-11-27 à 07:47, raja a écrit :
>
>>
>> ya, i made the comparison within the pitchshifting-playground patch  
>> which jean-francois mentioned and chose the transposer object  
>> instead of gizmo~ for my own needs because the doppler technique it  
>> uses seems to use half the CPU of gizmo's FFT technique.  
>> Personally, if I'm pitch-shifting, I'm not really concerned with  
>> artifacts appearing while making radical shifts so gizmo~ is a bit  
>> of overkill for me.
>>
>> check it out:
>> Max5(or4.6)/examples/effects/transposer-folder/lib/transposer.maxpat
>
>


what do you mean, i can only give it two arguments?

> what do you mean, i can only give it two arguments?

gizmo~ cpu usage question