Forums > MaxMSP

banded noise

March 30, 2006 | 3:14 pm

I need to create a patch that will allow me to create tightly banded noise.

i vaguely know the theory. I need to use two channels of noise, on in inverse phase from one another and use a tight eq on one side, which should reproduce tightly banded noise of my selected EQ width.

can anyone help? i’ve said it before, i am pretty new to max/msp and dont really know which objects to use.

if anyone has any simple patchs that they think might help, my email is

psc.recordings@gmail.com

cheers


March 30, 2006 | 4:11 pm

On 30-Mar-2006, at 17:14, John wrote:
> I need to create a patch that will allow me to create tightly
> banded noise.

I’ve done work in this direction in w/Litter Power (of course).

The current status of this particular object can be described as
"advanced beta testing". A few people have worked with the external,
in fact a CD has just been released using it. But there are some
gotchas I want to iron out before unveiling it to the world.

Still, you find some useful things the Litter Power online
information. URI below.

Beset — Peter

————– http://www.bek.no/~pcastine/Litter/ ————-
Peter Castine +–> Litter Power & Litter Bundle for Jitter

iCE: Sequencing, Recording & |home | chez nous|
Interface Building for |bei uns | i nostri|
Max/MSP Extremely cool http://www.castine.de

http://www.dspaudio.com/


March 30, 2006 | 6:25 pm

John wrote:
> can anyone help? i’ve said it before, i am pretty new to flash and
> dont really know which objects to use.

I don’t know flash at all, but regarding Max, have a look at the
forbidden planet in the examples folder I think. Its fun with
(a)bandened noise…

Stefan

[][] [][][] [][] [][][]
[][][][][][][][][][][][][][][]

Stefan Tiedje
Klanggestalter
Electronic Composition
&
Improvisation

/~~~~~
\ /|() ()|
))))) )| | |( \
/// _/)/ )))))
___/ ///

————————-x—-
–_____———–|———–
–(_|_ —-|—–|—–()—-
– _|_)—-|—–()———–
———-()————x—–

14, Av. Pr. Franklin Roosevelt,
94320 Thiais, France
Phone at CCMIX +33-1-57 42 91 09



f.e
March 31, 2006 | 6:41 am

which one of the beta LP is it ?


March 31, 2006 | 7:26 am

Hi,

look at the rand~ help file…

All the best

Alessandro Fogar


March 31, 2006 | 10:50 am

I cnat get litter power, for i am a broke student. i can get the free one however, will this still help, do you think?


March 31, 2006 | 1:40 pm

On 31-Mar-2006, at 12:50, John wrote:
> i can get the free one however, will this still help, do you think?

The Starter Pack gets you brown and pink noise, triangular dithering
noise, and an ultra-efficient 1-bit noise source is coming in the
next release. Plus some highly useful random number generators and
much other stuff. Of this, the signal degrader (lp.nn~) and one
chaotic model (lp.poppy~) just might prove useful in your project.
But maybe not as direct a help as lp.epoisse~ (coming soon) or rand~.

On the whole, I would always recommend downloading the Starter Pack.-)

> I cnat get litter power, for i am a broke student.

I aimed to price Litter Pro affordably for both Starving Artists and
Impoverished Students. It’s less than $1 per external. Hope you can
rise from broke to merely impoverished soon. (Believe me, I have been
all of the above.)

Maybe your school should sign up for a lab license?

Best,
Peter


[Please cf the .sig from my previous post on this thread if you need it]


March 31, 2006 | 2:15 pm

i know i might sound like a complete rooky (and believe me, i am) but how can i achieve what i want with these objects?

would i simply use one noise object routed for two channels. oen with a delay line (delay~ ?) to knock it out of phase and an EG to get the band i want (object?)

or is there a more direct method?


March 31, 2006 | 2:47 pm

>i know i might sound like a complete rooky (and believe me, i am)
>but how can i achieve what i want with these objects?
>
>would i simply use one noise object routed for two channels. oen
>with a delay line (delay~ ?) to knock it out of phase and an EG to
>get the band i want (object?)
>
>or is there a more direct method?

well i did (and do) what you describe (if I understand it well) using
very narrow (or not so narrow, depending of the ambitus I am after)
resonant filters

cnmat’s resonators~ work very well

band pass filters are of course another possibility, those made by
FFT~ are great as well (brickwall)

is this what you need??

kasper

Kasper T. Toeplitz
noise, composition, bass, computer

http://www.sleazeArt.com


March 31, 2006 | 3:09 pm

yes. bascially i need to create brickwall banded strips of noise, so that i can map formants from a human voice onto them.

ideally, i want to create a system i can affect in real time

does this fft~ require the methodology i explained above? or does it work as an accruate ‘brickwall’ band EQ?


March 31, 2006 | 3:23 pm

< <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
The Princeton Laptop Orchestra (PLOrk)
with special guests Zakir Hussain, Pauline Oliveros, and So Percussion

directed by Dan Trueman and Perry Cook

Tuesday April 4, Richardson Auditorium, Princeton University, 8pm

The Princeton Laptop Orchestra (PLOrk) is a new ensemble of 15
laptopists, each seated on a pillow with a 6-channel hemispherical
speaker and a variety of control devices. This premiere performance
will include works by Paul Lansky, Brad Garton, Curtis Bahn and Tomie
Hahn, Dan Trueman, Scott Smallwood, Seth Cluett, Perry Cook and Ge
Wang, with special guest performances by renowned tabla virtuoso Zakir
Hussain, accordian legend Pauline Oliveros, and the hot percussion
quartet So Percussion (who will be processing Hussain in real-time).

Tickets $10, $6 students (free with Tiger tickets).
Call 609-258-5000 or visit http://www.princeton.edu/utickets

http://plork.cs.princeton.edu
< <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<


March 31, 2006 | 3:44 pm

>yes. bascially i need to create brickwall banded strips of noise, so
>that i can map formants from a human voice onto them.
>
>ideally, i want to create a system i can affect in real time
>
>does this fft~ require the methodology i explained above? or does it
>work as an accruate ‘brickwall’ band EQ?

once you filter i don’t really see the use of pahse reversal – it might be me

but for what you describe, please, see the help files for resonators~
(and other cnmat objects)

kasper


March 31, 2006 | 4:01 pm

my tutor claims that to obtain a truley brickwall noice, i need to phase it out then EQ one channel, which will isolate the EQed sound and remove everything else


March 31, 2006 | 6:26 pm

On 31-Mar-2006, at 18:01, John wrote:
> my tutor claims that to obtain a truley brickwall noice, i need to
> phase it out then EQ one channel, which will isolate the EQed sound
> and remove everything else

The Fourier approach can give you as "truly" a brick wall as
anything, but is limited to the bandwidths defined by the Fourier
Transform bins.

Lots of people have done this. Again, cf Forbidden Planet and play
with that.

The rand~ examples and lp.epoisse~ take another approach to a kind of
frequency-banded noise. I honestly don’t know if they’re what you’re
after. Don’t hurt none to try at least rand~.help.

The EQ thing you’re describing sounds buildable with a couple of MSP
objects (delay~, filters of your choice, etc.).

It’s probably time to get the tools cracking and see what you can build.

– P.

[Please cf the .sig from my previous post on this thread if you need it]


March 31, 2006 | 7:23 pm

Do a search for Trond Lossius’ (free) externals and look at [tl.crossover4~].

It’s a crossover, so two of these in series will provide a brick-wall bandpass.

If you can wait until tomorrow, I can post a patch I made that does it.

Adam


March 31, 2006 | 7:47 pm

Quote: 88mph wrote on Fri, 31 March 2006 08:09
—————————————————-
> yes. bascially i need to create brickwall banded strips of noise, so that i can map formants from a human voice onto them.
>
> ideally, i want to create a system i can affect in real time
>
> does this fft~ require the methodology i explained above? or does it work as an accruate ‘brickwall’ band EQ?
—————————————————-

well for this job fft would not be wrong.
opne of the examples in the example folder is
an fft eqqualizer .. you can easily (well..)
make up your analyzer from that.

but forget the idea of "phaselinear brickwall
equalizer", this job would be far too CPU
intensive for todays computers.
and you do not need it all all for your human
voices experiment.
voices will not result in big offsets (like 24 db)
in adjacent fft bands when you use, say, 128 bands.
what is more useful is making enough fft bands,
like 2048 or more.

-fft 110


March 31, 2006 | 9:45 pm

This has so far been a very interesting thread, and I’ve gleaned quite a lot
of useful information from everyone’s responses.

But every time I see this subject line in my Inbox, I invariably read it as
"Brandied Noise", which sounds oddly delicious.

:-


April 1, 2006 | 12:37 am


April 1, 2006 | 3:39 am

>
> If you can wait until tomorrow, I can post a patch I made that does it.
>
> Adam
—————————————————-

that would be fantastic. tomorrow (well, its 4am and i’ve just finished work) was my day earmarked to begin drawling through all the suggestions here

thanks to everyone so far, most enlightening thread, so proud i started it ;)



f.e
April 1, 2006 | 10:00 am

Coming back from your site, and wonder if you can say more about Chuck ?!

f.e


April 1, 2006 | 11:53 pm

Hi,

I guess tomorrow means tomorrow, not today. Sorry, got stuck on some other things. But by Sunday, for sure.

However, now that I understand better what you want to do, I’m not sure my patch will do it. tl.crossover4~ isn’t meant for realtime frequency shifting — Depending on the settings, it causes some noticable audio glitches when sweeping frequencies. Also, using two of them to get a band-pass filter is cpu intensive.

But, I’ll post the patch anyway.

Adam


April 2, 2006 | 12:05 am

I was able to post the patch and it’s supporting patches:

http://www.hellbender.org/programming/brickwall_bandpass.zip

Requires tl.crossover4~, but Trong Lossius.

bandpass_stereo~.mxt is the parent patch, start with that.

Adam


Viewing 22 posts - 1 through 22 (of 22 total)