Glottal Flow Derivative

Glottal Flow Derivative

A technique for producing waveforms in articulatory synthesis.

Gist is: Find some mathematical way to approximate the flow of air in the glottis. The derivative of this curve is the waveform suitable for a time domain audio signal.

Papers I am looking at that deal with this:

[gasrhedw]: Hui-Ling Lu Dissertation

[gfprlahl]: glottal source modelling for singing (Lu, Smith)

[glhfjiwq]: Fant 1985, Four Parameter Model (aka LF Model)

[greswfhd]: Fant 1995, AKA the Transformed LF Model

[goqfwkij]: Lin Dissertation on Articulatory Synthesis (1991), Lu references this paper so I grabbed it (it also has many other things).

[gpoiesop]: Strik, 1998. This paper (allegedly) approaches ways to automatically fit an LF model, given the timing parameters. Lu's Glottal Source Modelling casually mentions this paper in one section.

[grwuwllf]: Velhuis, 1998. Another alternative to the LF model, Called the Rosenberg++ Model, built on the Rosenberg Model (1971). Supposedly more efficient than LF? And hopefully easier to compute (not sure yet).

[guiupqfw]: Mushtaq and Khalid (2010). This paper outlines a physically based glottal flow model, which is different than the Fant analytical models. This seems to use a mass-and-spring model, which Perry Cook's Dissertation seems to touch on all the way in 1991. However, this one claims to be quite efficient compared to other techniques.