Browse Prior Art Database

Method for Text Annotation Play Utilizing a Multiplicity of Voices

IP.com Disclosure Number: IPCOM000104994D
Original Publication Date: 1993-Jun-01
Included in the Prior Art Database: 2005-Mar-19
Document File: 2 page(s) / 72K

Publishing Venue

IBM

Related People

Flores, D: AUTHOR [+4]

Abstract

A program is described which allows text to be associated to a "voice template" used for playback when converting text into speech.

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

Method for Text Annotation Play Utilizing a Multiplicity of Voices

      A program is described which allows text to be associated to a
"voice template" used for playback when converting text into speech.

      The program described herein is composed of various processes.
A process called the "Voice/Text Binding process" lets the user, via
a graphical end-user interface, highlight fragments of text from a
file and associate them to a "voice profile".  The voice profile
contains instructions in an object code form that defines the
properties of an artificially generated human voice.  For example, a
voice profile may define the sound characteristics of the voice of a
young girl; another profile may define the voice of an old man, etc.
A user could associate some sentences of the text to the voice
profile of the young girl, and others to the profile of the old man.
When the text is converted into speech and played back, the sound
produced will be a computer-generated dialog between a young girl and
an old man.

      The voice profiles are created by a process called the "Voice
Profile Generator".  This process lets the user synthesize artificial
voices by specifying the values of the different properties that
comprise a speech waveform.  The end user interface of this process
may consist of a panel where the user can control the timbre,
frequency range, etc. of the voice to be defined.  By varying these
voice-defining values, the user can modify a sample voice profile to
generate voices for different individuals.

      The output of the Voice/Text Binding process is an intermediate
file which correlates strings of text to voice profiles.  A
Generalized Data Stream (GDS) architecture is a natural embodiment
for the implementation of this output.  The file will consist of
records for each fragment of text that is associated to a voice
profile.  These records will contain a fiel...