Text analysis system with letter sequence recognition and speech stress assignment arrangement

• US 4,829,580 A
• Filed: 03/26/1986
• Issued: 05/09/1989
• Est. Priority Date: 03/26/1986
• Status: Expired due to Term

• Current Assignee: American Telephone & Telegraph Company (AT&T, Inc.), Bell Telephone Laboratories, Inc. (Nokia Corporation)
• Original Assignee: American Telephone & Telegraph Company (AT&T, Inc.)
• Inventors: Church, Kenneth W.
• Primary Examiner(s): Harkcom, Gary V.
• Assistant Examiner(s): Anderson, Lawrence E.
• Application Number:
  US06/844,077
• Time in Patent Office: 1,140 Days
• Field of Search: 364/200 MS File, 364/900 MS File, 364/300, 364/513.5, 364/419, 381/41, 381/43, 381/52
• US Class Current: 704/260
• CPC Class Codes: G06F 40/216 using statistical methods
• CPC Class Codes: G06F 40/263 Language identification
• CPC Class Codes: G10L 13/08 Text analysis or generation of parameters for speech synthesis out of text, e.g. grapheme to phoneme translation, prosody generation or stress or intonation determination

Abstract

In text-to-speech generating arrangements, text is analyzed to determine the language source of words therein by successively selecting letter sequences of each word in the text. At least one signal representative of the probability that the text word corresponds to a particular language source is generated responsive to the selected letter sequences and a language source is selected for converting the text work to a phonetic string responsive to the probability representative signals.

Claims

1. A method for analyzing text having words from a plurality of language sources comprising the steps of:

generating signals representative of successive letter sequences of a word in the text;

forming at least one signal representative of the probability that each of the successive letter sequences corresponds to a particular language source, including;

storing a set of signals, each signal corresponding to the probability of occurrence of a prescribed letter sequence in a particular language;

addressing said stored probability of occurrence signal responsive to each successive letter sequence; and

generating a signal representative of the probability of occurrence of said text word in said particular language responsive to said addressed letter sequence-probability of occurrence signals.

2. A method for analyzing text having words from a plurality of language sources according to claim 1 wherein each letter sequence probability of occurrence signal corresponds to the conditional probability of the occurrence of the last letter of said sequence given the occurrence of the other letters of said sequence.

3. A method for analyzing text having words from a plurality of language sources according to claim 2 wherein said word probability signal forming step further comprises:

generating a signal representative of the probability of the text word being in any one of said particular languages; and

dividing the signal representative of the probability of occurrence of said text word in said particular language by the signal representative of the probability of the text word being in any one of said particular languages to form a probability estimate signal; and

forming a signal representative of the confidence level of said probability estimate signal responsive to the probability estimate signals.

4. A method for analyzing text of the type claimed in claim 1 and for generating synthesized speech from the analyzed text, further comprising the steps of

generating speech from the analyzed text;

including

generating a stress pattern for each analyzed text word responsive to the signal representative of the probability of occurrence of said text word in said particular language.

5. Apparatus for analyzing text having words from a plurality of language sources comprising:

means for generating signals representative of successive letter sequences of a word in the text;

means responsive to the generating means for forming at least one signal representative of the probability that the text word corresponds to a particular language source, including;

means for storing a set of signals each signal corresponding to the probability of occurrence of a prescribed letter sequence in a particular language;

means responsive to each successive letter sequence for addressing said stored probability of occurrence signal; and

means responsive to said addressed letter sequence probability of occurrence signals for generating a signal representative of the probability of occurrence of said text word in said particular language.

6. Apparatus for analyzing text having words from a plurality of language sources according to claim 5 wherein each letter sequence probability of occurrence signal generating means comprises means for forming a signal representative of the conditional probability of the occurrence of the last letter of said sequence given the occurrence of the other letters of said sequence.

7. Apparatus for analyzing text having words from a plurality of language sources according to claim 6 wherein said word probability signal forming means further comprises:

means responsive to said word probability of occurrence signals for the particular languages for generating a signal representative of the probability of the text word being in any one of said particular languages;

means for dividing the signal representative of the probability of occurrence of said text word in said particular language by signal representative of the probability of the text word being in any one of said particular languages to form a probability estimate signal; and

means responsive to the probability estimate signals for forming a signal representative of the confidence level of said probability estimate signal.

8. Apparatus for analyzing text of the type claimed in claim 5 and for generating speech from the analyzed text, further comprising

means for generating speech from the analyzed text, in which speech the stress pattern is responsive to the signal representative of the probability of occurrence of said text word in said particular language.