Written language parser system

• US 5,157,759 A
• Filed: 06/28/1990
• Issued: 10/20/1992
• Est. Priority Date: 06/28/1990
• Status: Expired due to Term

• Current Assignee: American Telephone & Telegraph Company (AT&T, Inc.)
• Original Assignee: AT&T, Inc.
• Inventors: Bachenko, Joan C.
• Primary Examiner(s): Fleming, Michael R.
• Assistant Examiner(s): Knepper, David D.
• Application Number:
  US07/546,127
• Time in Patent Office: 845 Days
• Field of Search: 381/44, 381/51-53, 395/2, 364/419
• US Class Current: 704/266
• 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
• CPC Class Codes: H04M 3/4938 comprising a voice browser which renders and interprets, e.g. VoiceXML

Abstract

An enhanced text-to-speech synthesizer accepts freely generated text sequences of words and synthesizes the received sequences with proper emphasis and with properly placed pauses. In combination with other elements, the synthesizer provides for an enhanced Dual Party Relay Service where the text generated by the sound-impaired party is synthesized without an attendant'"'"'s intervention. The text generated by users is made more intelligible by interpreting abbreviations, correcting errors (misspellings and "noise"), translating special terms that are used by the community of users, de-emphasizing words based on syntactic considerations and inserting pauses to enhance intelligibility.

Claims

1. A converter for synthesizing a speech signal comprising:

a word detector responsive to a freely generated text signal for detecting individual words in said text signal and developing thereby a string of words to be synthesized;

categorizing means for analyzing each word in the string of words with reference to N neighboring words in the string of words, where N is an integer, to determine the category to which each of the analyzed words belongs; and

syntax augmenting means responsive to said categorizing means for considering each word in the string and, in the process of considering each word, for inserting a pause generation signal in the string of words, before or after the considered word, when appropriate, based on the category of the considered word.

2. The converter of claim 1 wherein said syntax augmenting means inserts the pause generation signal before or after the considered word, when appropriate, based on the considered word'"'"'s category and the category of the one of the words immediately adjacent to the considered word.

3. The converter of claim 1 wherein said syntax augmenting means inserts the pause generation signal before or after the considered word, when appropriate, based on the considered word'"'"'s category and the category of the words immediately adjacent on both sides of the considered word.

4. The converter of claim 1 wherein said syntax augmenting means, additionally, associates a de-emphasis generation signal with each considered word, when appropriate, based on the category of the considered word and on the categories of words immediately adjacent to the considered word.

5. The converter of claim 4, further including a speech synthesizer, responsive to words detected by said word detector, to said pause generation signals and to said de-emphasis signals, for developing a speech signal.

6. The converter of claim 1 wherein said categorizing means includes an N word shift register for passing words through said shift register to form an N word sliding window for analyzing the words in said string of words received by said categorizing means.

7. The converter of claim 1 wherein said categorizing means includes word storage means and word access means for creating an N word window that slides through said string of words to enable analysis of the words received by said categorizing means.

8. The converter of claim 1, further including a speech synthesizer, responsive to the string of words augmented with pause generation signals by said syntax augmenting means, for developing a speech signal.

9. The converter of claim 8, further including an output FIFO memory interposed between said synthesizer and said syntax augmentation means.

10. The converter of claim 8 wherein said output FIFO memory delivers signals to said synthesizer under control of said pause generating signals.

11. The converter of claim 8, having the output signal of said synthesizer connected to a communications network for sending speech signals to said communications network.

12. The converter of claim 11, further comprising a speech-to-text converter connected to said communications network for receiving speech signals and converting received speech signals to text.

13. The converter of claim 1, further including a FIFO memory interposed between said word detector and said categorizing means for storing words detected by said word detector.

14. The converter of claim 13, wherein said FIFO memory stores words under control of said word detector and outputs words under control of said categorizing means.

15. The converter of claim 1, further including an output FIFO memory responsive to said syntax augmentation means.

16. The converter of claim 1 further comprising word modification means within said word categorizing means for recognizing words that are abbreviated forms of full words and for replacing the abbreviated forms with their corresponding full words.

17. The converter of claim 1 further comprising word modification means within said word categorizing means for recognizing and deleting words that are noise words and word portions that are noise portions.

18. The converter of claim 1 further comprising word modification means within said word categorizing means for recognizing the absence of words that are necessary for a chosen degree of intelligibility and injecting the missing words.

19. The converter of claim 1 further comprising word modification means within said word categorizing means for recognizing word phrases and categorizing each word phrase as a single unit.

20. The converter of claim 1 further comprising word modification means within said word categorizing means for recognizing misspelled words and for replacing said misspelled words with their respective correctly spelled words.

21. The converter of claim 1 further comprising word modification means within said word categorizing means for recognizing a predetermined sequence of words and for replacing said sequence of words with another sequence of words.

22. The converter of claim 1 further comprising word modification means within said word categorizing means for recognizing a word in said string of words as being in improper form and replacing the word having said improper form with a corresponding word having the proper form.

23. The converter of claim 1 further comprising word modification means within said word categorizing means for recognizing the detected individual word as a combination of words and for separating said words from the combination to result in a multiple-word phrase.

24. The converter of claim 1 further comprising word modification means within said word categorizing means for modifying a sequence of words in accordance with a set of syntactic rules customized for a particular set of users.

25. The converter of claim 1 further comprising word modification means within said word categorizing means for modifying a sequence of words in accordance with a set of standard English syntactic rules.

26. A converter for synthesizing a speech signal comprising:

a word detector responsive to a freely generated text signal for detecting individual words in said text signal and developing thereby a string of words to be synthesized, said string of words forming a set;

word modification means for transforming, in accordance with pre-established rules, said set formed by the detected string of words into a different set having another string of words;

categorizing means, responsive to the other string of words from said different set, for analyzing each word in said other string of words with reference to N neighboring words in said other string of words, where N is an integer, and for determining thereby the categories to which each of the analyzed words belongs,

syntax augmenting means responsive to said categorizing means for identifying pause-accepting words in said other string of words based on said categories and for inserting pause generation signals, in association with the identified words, into said other string of words to be synthesized.

27. The invention of claim 26 wherein said word modifying means recognizes words in said string of words that are noise words and deletes the noise words from said set to form said other string of words in said different set.