I had a go a number of years ago at coding an LPC-based formant detector for PD, based on the Speech Filing System from University College London. It has since remained on my "to do" list, since I got the object to load but there were problems in getting it working (i.e. it didn't, and crashed PD).
In theory, if you can detect the peaks in a Linear Predictive Coding-derive polynomial plot, then you can determine the vowel very accurately.
It would be nice to make this work - I'll dig it out and perhaps someone can help me understand why it doesn't work (and fix it). I'll dig it out and try to GDB it tonight, but I have to feed my child now.
Best, Ed Metastudio 4 for Pure Data - coming soon! Metastudio 3 still available at http://sharktracks.co.uk/puredata
----- Original Message ---- From: Mathieu Bouchard matju@artengine.ca To: Ludwig Maes ludwig.maes@gmail.com Cc: Pd List pd-list@iem.at Sent: Thu, 28 October, 2010 15:26:05 Subject: Re: [PD] IPA Vowel space analyzer
On Thu, 28 Oct 2010, Mathieu Bouchard wrote:
Then for example I numbered the above like 001 = 6 0 1 = [ɶ] 200 = 4 0 0 = [ɛ] 511 = 1 1 1 = [ʏ] 531 = 5 3 1 = [ʊ]
sorry, I screwed that one. I had gotten the first digit backwards, and I began to fix it and didn't finish. The correct list is :
001 = 0 0 1 = [ɶ] 200 = 2 0 0 = [ɛ] 511 = 5 1 1 = [ʏ] 531 = 5 3 1 = [ʊ]
else it wouldn't be consistent at all.
| Mathieu Bouchard ------------------------------ Villeray, Montréal, QC