Forums > MaxMSP

pfft~ tut 26 question

May 12, 2008 | 6:19 am

Hi

in tutorial 26, the last example about pfft~ (mypvoc) uses a buffer~inside the pfft~patch to record spectral data and play it back at variable rate without changing pitch, using a phasor.

So far so good.

But now I’m wondering why, if i try to save this buffer~and reload it later, it doesn’t sound the same at all. I tried either to save/load it inside or outside the pfft∞, but with the same disappointing results.

I suppose this has something to do with the "multiband" architecture of the pfft~ subpatch ?

Any ideas ?

thanks !


May 12, 2008 | 1:17 pm

To save/recall, you just need that, right?

– Pasted Max Patch, click to expand. –

Can you describe in which way it doesn’t seem the same "at all"?
Are frequency bins shuffled?
Or does it look like the original with a lesser quality?
J-F.

> But now I’m wondering why, if i try to save this buffer~and reload it later,
> it doesn’t sound the same at all. I tried either to save/load it inside or
> outside the pfft∞, but with the same disappointing results.


May 12, 2008 | 3:55 pm

Thanks, indeed, i suppose i just need that.

But i tried this inside and outside the pfft~subpatch, with the same poor result.

The sound seems a lot less detailed, as if a lot of frequencies were missing or something, difficult to describe.

Maybe I’m missing something


May 12, 2008 | 4:32 pm

On 12 May 2008, at 17:55, Philippe Jelli wrote:
>
> Thanks, indeed, i suppose i just need that.
>
> But i tried this inside and outside the pfft~subpatch, with the
> same poor result.
>
> The sound seems a lot less detailed, as if a lot of frequencies
> were missing or something, difficult to describe.
>
> Maybe I’m missing something

writing buffer contents to disk defaults to 16bit int data, afaik.
upon load this will be converted back to 32 bit floats but mapped to
-1 < --> +1 range.
not very useful in your situation.
send a "samptype float32" message to buffer, then "write".
volker.


May 12, 2008 | 5:56 pm

>> But i tried this inside and outside the pfft~subpatch, with the
>> same poor result.
>>
>> The sound seems a lot less detailed, as if a lot of frequencies
>> were missing or something, difficult to describe.

> writing buffer contents to disk defaults to 16bit int data, afaik.
> upon load this will be converted back to 32 bit floats but mapped to
> -1 < --> +1 range.
> not very useful in your situation.
> send a "samptype float32" message to buffer, then "write".
> volker.

Another possibility is to use a jit.matrix with type float32 instead of a
buffer.
J-F.


May 12, 2008 | 9:50 pm

Thanks for these advices. will try to see which works best.

Thus, by definition, if i hear you correctly, it should be possible to save and reload this buffer~ either inside or outside the pfft~ subpatcher ? (with the limitations in quality you spoke about)

Philippe


May 13, 2008 | 1:13 pm

The buffer is a place in the computer’s memory.
It is neither "inside" or "outside" of the pfft~ subpatcher.
Whether you have several [buffer~] objects with the same name, inside or
outside subpatchers, makes no difference soundwise. You can use the "read"
and "write" messages anywhere in the patch, it won’t make a difference.
J-F.

http://www.jeanfrancoischarles.com
May 12th update: Helianthus musca and Helmut Lachenmann

>
> Thus, by definition, if i hear you correctly, it should be possible to save
> and reload this buffer~ either inside or outside the pfft~ subpatcher ? (with
> the limitations in quality you spoke about)
>
> Philippe


May 13, 2008 | 1:59 pm

the samptype float32 message did the trick.

It now sounds as it should, difference was huge in my case;

Thanks again !

Philippe


Viewing 8 posts - 1 through 8 (of 8 total)