|| Martin Rothenberg – Maybe 20 or so years
ago, we were working on that problem, in the Syracuse
research lab, with some of the people here in the room. We tried to stimulate
the vocal tract using an impulse, which is a broadband source, and
we had an assortment of firearms. I think we had some success with
air pistols, but we didn’t have the sophisticated processing
algorithms we have now. Added to your methods for determining the
resonance would be inverse filtering […..listen to the audio link
for the rest of the discussion] Audio
||Jim Doing – I have a question about "what note is
this." Are these all the different notes, the dots? ….
I think it’s a real problem, I know, trying to find really
qualified singers to do these studies on, especially tenors since
so few tenors can sing to the top. …[listen to audio link
for remaining discussion] Audio
||Sten Ternström – I think your concern over calling
the resonances R 1, 2, 3 instead of F 1, 2, 3 is perhaps overly
cautious. In my mind, a formant is what you call a resonance, and
what you call a formant is something that I’d rather
not invoke at all. I think what you are saying is when we look
at spectrum section there are peaks and valleys and we tend to call
those peaks the formants, but in fact they are only the shadow,
or the footprint of the formants that are really the resonances in the vocal tract [….listen
to the audio link for remaining discussion.] Audio
||Johan Sundberg – Your results are in very good agreement
with what we were doing with much more coarse methods previously.
Even the rise of the 3rd and 4th formant I think was even in agreement.
What I think is important is that the first formant is always slightly
higher than the fundamental. When you looked at the tenors,
did you find that they adhered to this principle of keeping the
first formant higher than the formant in all vowels, or were
there exceptions? Audio link.