Re: Variability of natural speech production ("James D. Miller" )

Subject: Re: Variability of natural speech production
From:    "James D. Miller"  <jamdmill@xxxxxxxx>
Date:    Fri, 8 Aug 2008 09:17:28 -0400

Dear Dr. Schmitt, It is well established that the acoustic information used by a listener to identify a consonant or a vowel is overlapping and distributed acoustically across a considerable span of time. To slightly paraphrase Alvin Libermann, "Acoustically, speech sounds (phones) do not come along like beads on a string." For example, the acoustic information used to identify a /d/ may start sometime in a preceding vocalic portion of the signal and may extend well into a following vocalic portion of the signal. Even though some have tried to identify the acoustic locations of consonants and vowels in running speech, they are, for the most part, misguided in that attempt. Even the simple idea that vowels are associated with a "steady-state" portion of the acoustic signal or with the "middle of the vocalic portion," are very often incorrect. Unfortunately, one has to study the dynamic acoustic characteristics of speech over a considerable time span and understand how perception is related to those dynamics. We are only beginning to know how to do that. Until that time, I fear that you will have to narrow your question to restrict it to an answerable one. James D. Miller James D. Miller, Ph.D. Principal Scientist Communication Disorders Technology, Inc. Indiana University Research Park 501 N. Morton Street Suite 215 Bloomington, IN 47404 Business Phone: (812)336-1766 Cingular Cell Phone: (812)360-0612 Quoting Nicola Schmitt <nicola.schmitt@xxxxxxxx>: > Dear List, > > We would like to know more about the acoustic properties of natural speech? > > Especialy we are interested in the distribution of spectral levels of > single phonemes, especially /f/, /s/, /sh/, /t/, /d/ /k/, /g/, /p/, > /b/ on two dimensions: voice effort (whispering .. talking .. > shouting) and speaker "clarity" (barely understandable ... clearly > understandable). Long-term spectra are not helping us. > > Is any one working on either of this topics? Or knows topic related > literature? > > Thanks for your answer in advance > > Best Regards > Nicola Schmitt > > Research Audiologist > > > -- > Ist Ihr Browser Vista-kompatibel? Jetzt die neuesten > Browser-Versionen downloaden: > --

This message came from the mail archive
maintained by:
DAn Ellis <>
Electrical Engineering Dept., Columbia University