thanks everybody.. this goes by far beyond my capabilities, so i decided to turn it around and create a sample-based patch. the basic idea is to control a light installation through sound. 60 neon tubes are controlled via a custom built max patch. i want to connect the brightness of these tubes to the sound of breathing. my first thought was that it would be easier to deduct data from a patch based on synthesis. after the information i gathered from the list i realize that this is an immense task that has its focus only next to my intention. so i decided to create a sample based patch. at the moment i am using [fiddle] to read the amplitude of the recorded breath sounds.. however there are some troubles, like its very, very jumpy.. it is also difficult to find a sonically equivalent parameter to "breath in.. breath out".. pitch doesnt work.. [env] only reads the RMS so it does not change too much.. amplitude is close to what we hear as rising and falling, but still very far from the natural perception.. maybe someone has an idea ? anyways.. i will keep you updated.. : )
alexandre d.
On 10/07/07, Thomas Grill gr@grrrr.org wrote:
And if you really want to go into details, beyond PD, you could try PRAAT. http://www.fon.hum.uva.nl/praat/
greetings, Thomas
Andy Farnell schrieb:
Breathing could be as easy as pushing a bit of rising and falling noise through some parallel bandpass filters.
To model the human vocal tract, mouth and lungs, and the air pressure at all places inside is the tricky bit to make it sound convincing. Is it for singing or a synthesised instrument like flute?.. because for playing an instrument there is forced breathing where the lips are more closed and that needs a slightly different model. Open breathing is more related to singing synthesis, two deep sources to mine are this thesis
http://www.cs.princeton.edu/~prc/SingingSynth.html
and this one
http://web.media.mit.edu/~moo/thesis/YEK_thesis.pdf
What you probably are *not* interested in are fof, paf, vosim type methods, which are more for voiced sounds. One that would be good for breathing is vocoder (filter bank). If you have a good tract model it's gonna be good for breathing, replace the voiced oscillator with different kinds of noise. There's a pvoc example in the Pd help files. Take snapshots for "oooh" and "aah" formants and try those filters with noise. To make it efficient crossfading between a few fixed filters might be okay. I haven't researched it much further, but you'd suppose different forces are at work breathing in and out, so they will be different sounds with different noise/pressure curves.
On Tue, 10 Jul 2007 01:13:40 +0200 "alexandre r. decoupigny" alex.decoupigny@gmail.com wrote:
hello there...
i want to build a patch that synthesizes the sound of breathing.. i know some pd and some synthesis.. any idea where and how to start ?
best
alexandre
alexandre r. decoupigny
www.listeningart.blogspot.com www.myspace.com/alexandredecoupigny Adore Apples. Love Green Apples. http://www.greenpeace.org/apple/