Forums > MaxMSP

Non – realtime (offline) FFT-IFFT options?

January 10, 2008 | 12:41 pm

Hi Everybody,

I am looking for a way to perform a non-realtime FFT/IFFT, I know how to do the realtime stuff with pfft~ etc. but I want to perform a FFT on a buffer~ wich is not longer than the FFT size. So basically I want to use it for analysing a buffer and I can’t see how to do this with the realtime stuff. I’ve also searched maxobjects.com where I find a lot off realtime stuff, but no offline stuff.

Thanks a lot,
Kessito


January 10, 2008 | 2:04 pm

Hi Kessito,

This should be possible with FTM, from IRCAM, although I don’t know
offhand what the maximum FT size is.


Owen

Kessito wrote:
> Hi Everybody,
>
> I am looking for a way to perform a non-realtime FFT/IFFT, I know how
> to do the realtime stuff with pfft~ etc. but I want to perform a FFT
> on a buffer~ wich is not longer than the FFT size. So basically I
> want to use it for analysing a buffer and I can’t see how to do this
> with the realtime stuff. I’ve also searched maxobjects.com where I
> find a lot off realtime stuff, but no offline stuff.
>
>


January 10, 2008 | 2:48 pm

You can use soundhack for this and then read the result with csound~
external.

Peiman

On 10 Jan 2008, at 14:04, Owen Green wrote:

> Hi Kessito,
>
> This should be possible with FTM, from IRCAM, although I don’t know
> offhand what the maximum FT size is.
>
> —
> Owen
>
> Kessito wrote:
>> Hi Everybody,
>> I am looking for a way to perform a non-realtime FFT/IFFT, I know how
>> to do the realtime stuff with pfft~ etc. but I want to perform a FFT
>> on a buffer~ wich is not longer than the FFT size. So basically I
>> want to use it for analysing a buffer and I can’t see how to do this
>> with the realtime stuff. I’ve also searched maxobjects.com where I
>> find a lot off realtime stuff, but no offline stuff.
>> maxmsp mailing list maxmsp@cycling74.com http://www.cycling74.com/
>> mailman/listinfo/maxmsp
>
>


January 10, 2008 | 4:36 pm

Is there a reason you can’t use the same [pfft~] methods, with
appropriate modifications, on a buffer instead of the real-time input
stream?

On Jan 10, 2008, at 5:41 AM, Kessito wrote:

>
> Hi Everybody,
>
> I am looking for a way to perform a non-realtime FFT/IFFT, I know
> how to do the realtime stuff with pfft~ etc. but I want to perform
> a FFT on a buffer~ wich is not longer than the FFT size. So
> basically I want to use it for analysing a buffer and I can’t see
> how to do this with the realtime stuff. I’ve also searched
> maxobjects.com where I find a lot off realtime stuff, but no
> offline stuff.

—-
Steven M. Miller
Professor, Contemporary Music Program
College of Santa Fe

Home < http://pubweb.csf.edu/~smill>
SFIFEM <
http://sfifem.csf.edu>
Atrium Sound Space <
http://atrium.csf.edu>
OVOS <
http://pubweb.csf.edu/~smill/ovos.html>
CMP <
http://www.csf.edu/academics/contemporary_music/>


January 10, 2008 | 6:45 pm


January 12, 2008 | 12:11 am

Kessito wrote:
> Hi Everybody,
>
> I am looking for a way to perform a non-realtime FFT/IFFT, I know how to do the realtime stuff with pfft~ etc. but I want to perform a FFT on a buffer~ wich is not longer than the FFT size. So basically I want to use it for analysing a buffer and I can’t see how to do this with the realtime stuff. I’ve also searched maxobjects.com where I find a lot off realtime stuff, but no offline stuff.
>
> Thanks a lot,
> Kessito

there is an external library called vasp (information is here:
http://grrrr.org, download is there: http://grrrr.org/ext/beta/max)
which does non realtime processes on buffers.) threaded and non threaded.
marius.


January 14, 2008 | 8:02 pm

On 12 Jan 2008, at 01:11, marius schebella wrote:

> Kessito wrote:
>> Hi Everybody,
>> I am looking for a way to perform a non-realtime FFT/IFFT, I know
>> how to do the realtime stuff with pfft~ etc. but I want to perform
>> a FFT on a buffer~ wich is not longer than the FFT size. So
>> basically I want to use it for analysing a buffer and I can’t see
>> how to do this with the realtime stuff. I’ve also searched
>> maxobjects.com where I find a lot off realtime stuff, but no
>> offline stuff. Thanks a lot,
>> Kessito
>
> there is an external library called vasp (information is here:
> http://grrrr.org, download is there: http://grrrr.org/ext/beta/max)
> which does non realtime processes on buffers.) threaded and non
> threaded.
> marius.

also i have an external that does fft/ifft on buffers in non realtime
that i would be happy to share.
osx only.
vb



kjg
January 14, 2008 | 10:38 pm

since vasp seems to be completely undocumented for max/msp right now, I’m sure a lot of people could make good use of your object, Volker.
Would you consider putting a link to it on maxobjects.com?

Regards,
Klaas-Jan


January 15, 2008 | 12:20 pm

Thank you all for the replies, I will check out the vasp and ftm options. The reasons why I can’t do it with fft~/pfft~ are:
- It seems impossible to get a proper synchronisation between the buffer~ and the fft~ object (you can’t make a zero delay feedback)

-For this project I need FFT sizes of approx. 4 million points, to my knowledge the fft~ object has a maximum size of 16k.


January 15, 2008 | 12:26 pm

On 14 Jan 2008, at 23:38, Klaas-Jan Govaart wrote:

>
> since vasp seems to be completely undocumented for max/msp right
> now, I’m sure a lot of people could make good use of your object,
> Volker.
> Would you consider putting a link to it on maxobjects.com?

ok -
i wonder how well this works for others anyway.
i’ve hesitated to put my stuff on maxobjects.com, cause most of the
objects were made for a specific project or simply for
proof of concept and therefore haven’t been tested very carefully, yet.
but since this state might not change "in the near future", i might
aswell just make them available as they are, and let
everyone else decide what is useful and what not.

will probably find some time tonight to upload. stay tuned.
volker.


January 15, 2008 | 2:36 pm


January 15, 2008 | 8:11 pm

No, I’m afraid that I really do need this 0,01 Hz resolution, this is because I’m doing some very low-frequency analysing according to
type 0 (lab) ISO specifications… I know that I can just lower the sample rate but I need the whole spectrum, and I could do split frames (different sample rates for different parts of the spectrum) but then things might get dodgy when I FFTconvolve stuff.

What I actually need to do is:
-record log freq sweep (couple of minutes)
-convolve with inverse sweep ->this gives the impulse response
-filter with passband filters
-make energy decay curves

btw, maybe someone has actually used vasp, and could tell just how to perform a FFT/IFFT with it?


January 15, 2008 | 10:39 pm

Hi Kessito,

You might find this helpful:

http://www.notam02.no/notam02/prod-prg-mammut-e.html


Owen

Kessito wrote:
> No, I’m afraid that I really do need this 0,01 Hz resolution, this is
> because I’m doing some very low-frequency analysing according to type
> 0 (lab) ISO specifications… I know that I can just lower the sample
> rate but I need the whole spectrum, and I could do split frames
> (different sample rates for different parts of the spectrum) but then
> things might get dodgy when I FFTconvolve stuff.
>
> What I actually need to do is: -record log freq sweep (couple of
> minutes) -convolve with inverse sweep ->this gives the impulse
> response -filter with passband filters -make energy decay curves
>
> btw, maybe someone has actually used vasp, and could tell just how to
> perform a FFT/IFFT with it?
>
>


January 16, 2008 | 12:12 am


January 16, 2008 | 2:49 pm


January 16, 2008 | 3:24 pm


January 17, 2008 | 5:20 pm


Viewing 17 posts - 1 through 17 (of 17 total)