Question about Vocal Tract Modeling

[Tarun Pruthi <tpruthi@xxxxxxxxxxxx>]

Hi all,

I have been working on vocal tract modeling for nasalized vowels using the
computer simulation program (VTAR) developed in our lab. It is usually
said in these cases that the poles of the system occur when

Bp + Bn + Bo = 0 where

Bp, Bn and Bo are the susceptances for the pharyngeal cavity (Bp), oral
cavity (Bo) and nasal cavity (Bn) (from the branching location). I am
using detailed vocal tract areas from MRI recordings.  During my
simulations I have observed that the zero crossings of Bp+Bo+Bn do not
correspond to the resonances observed in the vocal tract transfer function
(computed by the same computer program). However, the resonances
correspond exactly to the zero crossings of Bi (where Bi is the total
susceptance of the vocal tract, looking into the vocal tract at the
glottis).

My question is: Should the zero crossings of Bp+Bo+Bn correspond to the
resonances of the transfer function when I am using the transmission line
model with multiple short tubes (modeling the detailed area function)? Or
is this valid only when the pharyngeal, oral and nasal cavities are
modeled as single uniform tubes? (My simulations show that the zeros of
Bp+Bo+Bn and the resonances of the transfer function do correspond to each
other when I model the cavities as simple tubes).

Is there a reference proof somewhere which proves that the zero crossings
of Bp+Bo+Bn should give the resonances of the transfer function even when
I am using a detailed area model? Or if it is not true is there a
reference which proves it the other way?

I am just wondering whether there is something wrong with the program
(although I have checked hard and don't seem to find anything wrong) or is
it theoretically incorrect to say that?

Thanks in advance,
Tarun

-----------------------------------------
Tarun Pruthi
Graduate Research Assistant, ECE
Room 3180, A V Williams Building
University of Maryland, College Park
MD 20742 USA
Email: tpruthi@glue.umd.edu
Web: www.ece.umd.edu/~tpruthi
Ph: 301-405-1365
----------------------------------------