Browse Prior Art Database

Tutorial Mode in a Speech Acquisition System

IP.com Disclosure Number: IPCOM000100316D
Original Publication Date: 1990-Apr-01
Included in the Prior Art Database: 2005-Mar-15
Document File: 2 page(s) / 66K

Publishing Venue

IBM

Related People

Epstein, ME: AUTHOR

Abstract

Disclosed is a tutorial imbedded in a speech acquisition system used to capture isolated-word utterances for speaker-dependent training of the Tangora (1).

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Tutorial Mode in a Speech Acquisition System

       Disclosed is a tutorial imbedded in a speech acquisition
system used to capture isolated-word utterances for speaker-dependent
training of the Tangora (1).

      In capturing training utterances for an isolated-speech
recognition system, it is beneficial to record sentences instead of
words.  This gives the system broad acoustic variations due to
coarticulation.  The problem is that new users of the system do not
know how to speak in isolated speech.  There are numerous problems
encountered by first-time users:
  They tend to run words together or syllabify words.
  Acronyms like "IBM" are confusing because users do not know if this
should be pronounced as one word or as the letters "I-B-M".
  The Tangora performs better when the acoustics for each sound are
pronounced in a consistent, i.e., monotonic way, but users do not
know the "ideal" way to speak to the Tangora.

      We have found that including an audio tutorial in the speech
acquisition system gives users sufficient preparation in both the
operation of the program and how to pronounce the training sentences.

      The tutorial contains numerous sentences that explain both the
operation of the system and how to speak in isolated speech.  When
the user runs the tutorial, each sentence is displayed, and a
recording for the sentence is played for the user.  Doing this gives
the user both a visual and audio version of the tutorial, and
demonstrates isolated- speech to the user.

      Another advantage of the tutorial is that it displays the
"bouncing-ball".  The bouncing-ball i...