Hello,

I was wondering, does the pvsfread opcode work in SDFT mode? And what
about the pvsblur, pvsmooth and pvstencil opcodes?

Thanks!

Hector

Hello again,

I was actually trying to create a pvx file with pvanal using a
hop-size of 1 (-n 2048 -h 1) and it seems that its not accepted as a
valid value since I get this:

pvanal: 2048 is a bad window overlap index

Looking at pvanal's source it seems that it expects a value of overlap
(frameSize/frameIncr) between 2 and 64. Does this mean that pvanal is
not SDFT ready? Will it be in the future?

Thanks,

Hector


On Tue, Mar 3, 2009 at 5:56 PM, Hector Centeno  wrote:
> Hello,
>
> I was wondering, does the pvsfread opcode work in SDFT mode? And what
> about the pvsblur, pvsmooth and pvstencil opcodes?
>
> Thanks!
>
> Hector
>

It was never my intention or expectation to support files with SDFT. It 
is conceive purely for streaming. At the very least, you are asking to 
create an analysis file with (say) 44100 frames per second. That 
compares with perhaps 172 fps for a normal analysis file. A doubles 
amp/freq frame size 2048 is 32768 bytes, one second at sr=44100 would 
take that to almost 1.5 GBytes.  The analysis file format is still based 
on the 32bit WAVE format, which is fine under normal circumstances, but 
will run out of capacity (together with Csound itself) very quickly 
indeed under such usage!

John Fitch added the SDFT option to Csound (almost certainly ~not~ to 
pvanal), and is best placed to to answer to what extent the other 
opcodes you mention support it. On the face of it, if what you want to 
do is blur the spectrum anyway, SDFT would be both overkill and probably 
less effective (and a lot less efficient!) than a standard streaming 
pvoc. The special advantages of the SDFT would largely be irrelevant.

Richard Dobson.


Hector Centeno wrote:
> Hello again,
> 
> I was actually trying to create a pvx file with pvanal using a
> hop-size of 1 (-n 2048 -h 1) and it seems that its not accepted as a
> valid value since I get this:
> 
> pvanal: 2048 is a bad window overlap index
> 
> Looking at pvanal's source it seems that it expects a value of overlap
> (frameSize/frameIncr) between 2 and 64. Does this mean that pvanal is
> not SDFT ready? Will it be in the future?
> 
> Thanks,
> 
> Hector
> 
> 
> On Tue, Mar 3, 2009 at 5:56 PM, Hector Centeno  wrote:
>> Hello,
>>
>> I was wondering, does the pvsfread opcode work in SDFT mode? And what
>> about the pvsblur, pvsmooth and pvstencil opcodes?
>>
>> Thanks!
>>
>> Hector
>>
> 
> 
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"
> 
> 

Thanks for the reply. I did consider that the resulting analysis files
would be quite large but didn't get to calculate the real figures.
1.5Gigs per second seems big enough to forget about using pvsfread
since it would have to load the analysis file in memory. I included
pvsblur in the list because it's part of a chain of spectral processes
I included in a orchestra and I was wondering if it would work if I
ran it as SDFT. I've been curious about trying the FM modulation (as
shown here: http://dream.cs.bath.ac.uk/SDFT/ ) and the quality of time
stretching and frequency shifting/stretching that SDFT would give.

Cheers,

Hector


On Wed, Mar 4, 2009 at 4:45 AM, Richard Dobson
 wrote:
> It was never my intention or expectation to support files with SDFT. It is
> conceive purely for streaming. At the very least, you are asking to create
> an analysis file with (say) 44100 frames per second. That compares with
> perhaps 172 fps for a normal analysis file. A doubles amp/freq frame size
> 2048 is 32768 bytes, one second at sr=44100 would take that to almost 1.5
> GBytes.  The analysis file format is still based on the 32bit WAVE format,
> which is fine under normal circumstances, but will run out of capacity
> (together with Csound itself) very quickly indeed under such usage!
>
> John Fitch added the SDFT option to Csound (almost certainly ~not~ to
> pvanal), and is best placed to to answer to what extent the other opcodes
> you mention support it. On the face of it, if what you want to do is blur
> the spectrum anyway, SDFT would be both overkill and probably less effective
> (and a lot less efficient!) than a standard streaming pvoc. The special
> advantages of the SDFT would largely be irrelevant.
>
> Richard Dobson.
>
>
> Hector Centeno wrote:
>>
>> Hello again,
>>
>> I was actually trying to create a pvx file with pvanal using a
>> hop-size of 1 (-n 2048 -h 1) and it seems that its not accepted as a
>> valid value since I get this:
>>
>> pvanal: 2048 is a bad window overlap index
>>
>> Looking at pvanal's source it seems that it expects a value of overlap
>> (frameSize/frameIncr) between 2 and 64. Does this mean that pvanal is
>> not SDFT ready? Will it be in the future?
>>
>> Thanks,
>>
>> Hector
>>
>>
>> On Tue, Mar 3, 2009 at 5:56 PM, Hector Centeno  wrote:
>>>
>>> Hello,
>>>
>>> I was wondering, does the pvsfread opcode work in SDFT mode? And what
>>> about the pvsblur, pvsmooth and pvstencil opcodes?
>>>
>>> Thanks!
>>>
>>> Hector
>>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>> csound"
>>
>>
>
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
> csound"
>