Very true, with pconvolve you can mimick convolve, too, by
setting psize = impulsize. The only downside with small partitions
is that the process is more costly (as you approach time-domain
convolution, rather than a fast freq-domain process).

Victor
----- Original Message ----- 
From: "Richard Dobson" 
To: 
Sent: Thursday, September 11, 2008 12:25 AM
Subject: [Csnd] Re: Re: Re: opcode convolve


> Because the audio input is analysed on the fly. It might in principle be a 
> real-time audio stream, not coming from a file at all. The classic 
> application is reverb, using a pre-analysed impulse response of a hall, 
> tank, bathroom, whatever, and convolving the audio input coming from your 
> live recording. Of course the long latency in this case is a major 
> limitation, and the newer pconvolve opcode would be much preferred; apart 
> from using "partioned convolution" with very much reduced latency, it 
> works in exactly the same way.
>
> Richard Dobson
>
>
> Federico Vanni wrote:
>> ok, but...
>> why only one analysis instead of two?
>>
>> fv
>>
>> Il giorno 10/set/08, alle ore 21:48, victor ha scritto:
>>
>>> one of the analyses is done by the utility cvanal. The spectrum
>>> is then stored and is read by convolve.
>>>
>>> Victor
>>
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound" 

ok... many thanks for your help.
fv