[Clam-devel] [PATCH] CardinalVowel Control, doxygen documentation for formant extractor

Pau Arumi parumi at iua.upf.edu
Thu Aug 16 10:06:18 PDT 2007


Hi Abe,
The patch have been commited (rev 10715)
and a photo have been taken :)
http://iua-share.upf.edu/wikis/clam/index.php/Development_screenshots

Notice the cathegory "Speech (Experimental - GSoC)" in the
processing tree. These processings have been commited in
CLAM/plugins/speech. Take a look how (easily) they are registed
with metadata. You can expand the description, if you wish.

Do you think it will be easy to improve the vowel model? Does not 
sound very human, right?

Cheers
Pau

En/na abe kazemzadeh ha escrit:
> Hi All,
> 
> Here's a patch that I've been working on.  It takes an input between 0 
> and 1 and outputs two formant values (for use with the vowel resonator). 
>  The input represents the location of the vowel with respect to cardinal 
> vowels, eight evenly spaced vowels that start with iy (high front vowel) 
> -> ey -> eh -> ae (low front vowel) -> aa (low back vowel) -> ao -> ow 
> -> uw (high back vowel). 
>  It can be thought of as a mapping from a one dimensional control (slider 
> position) 
> to a 2D one (formants 1 and 2).  To visualise it, you can think of marking 8 vowels in the control surface, connecting them, and then projecting the 2d curve into 1d.   
> 
> 
> Some issues:
> * I had trouble getting 
> the class to be registered from CLAM/pluggins/speech, so I just compiled it in CLAM/src/Processing/Controls and registered it manually in NetworkEditor/src/ProcessingTree.cxx. 
>  Pau, do you know how I can do this in CLAM/pluggins/speech?  Let me 
> know and I can put it there.
> * right now I just manually fit a function for each formant.  It would 
> be cool to do this process automatically (the user could make custom 
> wah-wah petals -- between different vowels besides just uw and aa 
> -- by giving a series of vowels ).
> * Right now, I have a control for # of steps.  I'm planning to use this 
> to make discrete vowels instead of continuous blending as one moves the 
> input slider.
> 
> this patch is in the attached file:
> CardinalVowel.patch
> 
> the patch to register this with the NetworkEditor is:
> Add_CardinalVowel_to_ProcessingTree.patch
> 
> Here's a network to test it out with:
> experimentalVowelSynth_withCardinalVowelControl.clamnetwork
> experimentalVowelSynth_withCardinalVowelControl.clamnetwork.pos
> 
> I'm also including a small patch with doxygen documentation for the 
> FormantExtractor class:
> FormantExtractor_add_documentation.patch
> 
> Let me know if anyone has any feedback... Thanks,
> Abe
> 
> 
> ------------------------------------------------------------------------
> 
> The original MIME headers for this attachment are:
> Content-Type: text/plain; charset="us-ascii"
> MIME-Version: 1.0
> Content-Transfer-Encoding: 7bit
> Content-Disposition: inline
> 





More information about the clam-devel mailing list