Newsgroups: comp.speech
Path: pavo.csi.cam.ac.uk!doc.ic.ac.uk!uknet!pipex!howland.reston.ans.net!cs.utexas.edu!swrinde!elroy.jpl.nasa.gov!decwrl!koriel!sh.wide!wnoc-kyo-news!atrwide!atr-la!paul
From: paul@itl.atr.co.jp (Paul Taylor)
Subject: Re: What physical grounding do formants have?
In-Reply-To: blix@asimov.cs.uiuc.edu's message of Mon, 10 Jan 1994 21:53:14 GMT
Message-ID: <PAUL.94Jan11113534@as51.itl.atr.co.jp>
Sender: news@itl.atr.co.jp (USENET News System)
Nntp-Posting-Host: as51
Organization: ATR Interpreting Telecommunications Research Labs., Japan
References: <CJFCuw.124@cygnus.com> <CJFosr.IDo@cs.uiuc.edu>
Date: Tue, 11 Jan 1994 02:35:34 GMT
Lines: 44

In article <CJFosr.IDo@cs.uiuc.edu> blix@asimov.cs.uiuc.edu (Gunnar Blix) writes:

   crash@cygnus.com (Jason Molenda) writes:

   >One thing about which I was not clear -- is the sound for a particular
   >formant tied to a particular part of the vocal tract, i.e. the
   >frequencies in the first formant (F1) are made between the glottis and
   >velum?

I'm not an expert in this either, but I do know some basic principles.
If the vocal tract was a uniform tube (it approximates to this
for some schwa sounds), you can predict the resonances from 
the length of the tube. The resonance pattern is changed when
cavities are formed by the tongue.

Strictly speaking it is impossible to derive the tongue position from 
formant frequencies, as the same spectrum can be generated by a wide
variety of tongue positions. If you have the tongue forming two cavities
A and B of different sizes it doesnt matter which is first, ie 

	glottis A tongue B teeth

should sound the same as

 	glottis B tongue A teeth

I'm a bit unclear on the details, but its basically just simple lumped
parameter system theory, ie you can get the same resonance pattern
from a wide variety of electrical and mechanical systems too.

However, that said, speakers clearly have heavy constraints on how
they use their tongue. You will probably find that
the traditional literature which says that /i/ is spoken with
a front high tongue position corresponds to what most speakers do.

So, there basically exists a many-to-one mapping between tongue
positions and formants, in that for each tongue position there is a
single set of formants, but a set of formants can be generated from a
variety of tongue positions.

Paul Taylor
ATR, Japan.


