Unified messaging system with automatic language identification for text-to-speech conversion

• US 6,487,533 B2
• Filed: 01/10/2000
• Issued: 11/26/2002
• Est. Priority Date: 07/03/1997
• Status: Expired due to Term

• Current Assignee: Avaya Incorporated
• Original Assignee: Avaya Incorporated
• Inventors: Hyde-Thomson, Henry C. A., Liron, Roger
• Primary Examiner(s): Banks-Harold, Marsha D.
• Assistant Examiner(s): Lerner, Martin
• Application Number:
  US09/480,296
• Publication Number:
  US 20010056348A1
• Time in Patent Office: 1,051 Days
• Field of Search: 704/2, 704/3, 704/8, 704/260, 704/270, 704/277, 379/88.05, 379/88.06, 379/88.16
• US Class Current: 704/260
• 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: G10L 15/005 Language recognition
• CPC Class Codes: H04L 51/066 Format adaptation, e.g. format conversion or compression
• CPC Class Codes: H04M 2201/60 Medium conversion
• CPC Class Codes: H04M 3/5307 for recording messages comprising any combination of audio and non-audio components

Abstract

A unified messaging system includes a voice gateway server coupled to an electronics mail system and a private branch exchange (PBX). The voice gateway server provides voice messaging services to a set of subscribers. Within the voice gateway server, a trigraph analyzer sequentially examines 3-character combinations within a text message; determines occurrence frequencies for the character combinations; compares the occurrence frequencies with reference occurrence statistics modeled from text samples written in particular languages; and generates a language identifier and a likelihood value for the text message. Based upon the language identifier, a message inquiry unit selects an appropriate text-to-speech engine for converting the text message into computer-generated speech that is played to a subscriber.

Claims

1. A method of operating language-based conversion of a present text message into speech, the method comprising the following steps:

a. retrieving the present text message;

b. automatically generating a language identifier corresponding to the present text message, wherein the step of generating includes;

(1) examining a sequence of characters of the present text message;

(2) determining an actual frequency of occurrence of character combinations within the sequence of characters of the present text message; and

(3) matching the actual frequency of occurrence of the character combinations within the sequence of characters of the present text message with one of a plurality of corecurrence libraries;

c. converting the present text message directly into computer-generated speech in a language corresponding to the language identifier using a language-specific text-to-speech engine, wherein the language-specific text-to-speech engine is selected according to the language identifier; and

d. playing the computer generated speech to a subscriber.

2. The method as claimed in claim 1, wherein the step of converting is performed by the language-specific text-to-speech engine utilizing a phoneme library.

3. The method as claimed in claim 1, further comprising the following steps:

a. sensing a subsequent text message; and

b. repeating the steps of retrieving, generating, converting, and playing in response to the step of sensing.

4. The method as claimed in claim 1, wherein the step of matching further comprising the steps of:

a. comparing the actual frequency of occurrence with each of a plurality of reference frequencies wherein each of the plurality of reference frequencies corresponds to one of the plurality of corecurrence libraries; and

b. determining a best match between the actual frequency of occurrence and one of the plurality of reference frequencies.

5. The method as claimed in claim 1, wherein the step of examining comprises using a trigraph analyzer for inspecting the character combinations and wherein the character combinations comprise three consecutive characters within the sequence of characters.

6. The method as claimed in claim 1, wherein the sequence of characters is found in a first portion of the present text message.

7. The method as claimed in claim 1, wherein the step of matching further comprising the following steps:

a. comparing the actual frequency of occurrence with each of a plurality of reference frequencies wherein each of the plurality of reference frequencies corresponds to one of the plurality of corecurrence libraries; and

b. determining that a sufficient number of matches exist between the actual frequency of occurrence and one of the plurality of reference frequencies.

8. The method as claimed in claim 7, wherein the step of matching is performed when there is the sufficient number of matches between the actual frequency of occurrence and one of the plurality of reference frequencies.

9. The method as claimed in claim 7, further comprising the step of terminating the method when the sufficient number of matches does not exist.

10. A method of providing language-based conversion of an original text message into speech for a user comprising the following steps:

a. retrieving the original text message;

b. automatically generating a language identifier corresponding to the original text message, wherein the step of generating includes;

(1) examining a sequence of characters of the original text message;

(2) determining an actual frequency of occurrence of character combinations within the sequence of characters; and

(3) matching the frequency of occurrence of the character combinations with one of a plurality of corecurrence libraries;

c. automatically selecting an appropriate one text translator from a plurality of text translators, wherein each of the plurality of translators corresponds to one of a plurality of languages and the appropriate one text translator is selected based upon the language identifier;

d. translating the original text message into a translated text message in a user selected language in response to the appropriate text translator;

e. converting the translated text message into computer generated speech based upon the user selected language; and

f. playing the computer generated speech to the user.

11. The method as claimed in claim 10, further comprising the following step:

polling the user for the user selected language.

12. The method as claimed in claim 10, wherein the step of matching further comprising the following steps:

a. comparing the actual frequency of occurrence with each of a plurality of reference frequencies wherein each of the plurality of reference frequencies corresponds to one of the plurality of corecurrence libraries; and

b. determining that there is a sufficient number of matches between the actual frequency of occurrence and one of the plurality of reference frequencies.

13. The method as claimed in claim 10, wherein the step of matching further comprising the following steps:

a. comparing the actual frequency of occurrence with each of a plurality of reference frequencies wherein each of the plurality of reference frequencies corresponds to one of the plurality of corecurrence libraries; and

b. determining a best match between the actual frequency of occurrence and the plurality of reference frequencies.

14. A messaging system for converting a text message into computer generated speech, the system comprising:

a. means for storing the text message;

b. means for automatically generating a language identifier corresponding to the text message wherein the means for automatically generating is coupled to the means for storing, wherein the means for generating includes means for determining an actual frequency of occurrence of character combinations within a sequence of characters of the text message and means for comparing the actual frequency of occurrence of the character combinations within the sequence of characters with a plurality of reference frequencies wherein each reference frequency corresponds to a particular corecurrence library;

c. a plurality of text-to-speech engines coupled to the means for storing wherein each of the plurality of text-to-speech engines corresponds to one of a plurality of languages and an appropriate one text-to-speech engine based on the language identifier converts the text message into the computer generated speech; and

d. means for playing the computer-generated speech to a subscriber.

15. The system as claimed in claim 14, further comprises a phoneme library coupled to the text to speech engine for converting the text message into the computer generated speech.

16. The system as claimed in claim 14, wherein the means for automatically generating further comprises a trigraph analyzer to formulate an occurrence frequency of the text message based on examining a combination of three consecutive characters within the text message.

17. The messaging system as claimed in claim 14, wherein the means for comparing the actual frequency of occurrence of the character combinations within the sequence of characters of the text message further comprises means for determining a closest match between the actual frequency of occurrence of the character combinations and one of the plurality of reference frequencies.

18. The messaging system as claimed in claim 17 wherein:

a. the means for determining the actual frequency of occurrence of the character combinations further comprises;

1. means for dividing the sequence of characters of the present text messages into a plurality of sequential character sets, and

2. means for determining the actual frequency of occurrence further comprises determining a set of actual rates at which each of the plurality of sequential character sets occur in the sequence of characters of the text message; and

b. the means for comparing further comprises means for matching the set of actual rates with one of the plurality of corecurrence libraries.

19. The messaging system as claimed in claim 18, wherein the means for matching the set of actual rates includes:

a. means for comparing the set of actual rates at which each of the plurality of sequential character sets occur in the sequence of characters of the text message with a plurality of reference occurrence frequencies for the sequence of characters wherein each reference occurrence frequency corresponds to one of the plurality of corecurrence libraries; and

b. means for determining a closest match between the set of actual rates and one of the reference occurrence frequencies.

20. A voice messaging system for providing voice messaging services to a set of subscribers, the voice messaging system comprising:

a. means for retrieving a text message;

b. means for automatically generating a language identifier corresponding to the text message, wherein the means for generating includes means for comparing an actual frequency of occurrence of character combinations within a sequence of characters of the text message with a plurality of reference frequencies wherein each reference frequency corresponds to a particular corecurrence library;

c. means for converting the text message directly into computer-generated speech, wherein the means for converting uses a language-specific text-to-speech engine that is selected based upon the language identifier; and

d. means for playing the computer-generated speech to a subscriber.

21. The voice messaging system according to claim 20, further comprising a voice gateway server configured to be connected to a computer network and a Private Branch Exchange, wherein the voice gateway server facilitates the exchange of messages between the computer network and the Private Branch Exchange.

22. A method of operating language-based conversion of a present text message into speech, the method comprising the following steps:

a. retrieving the present text message;

b. automatically generating a language identifier corresponding to the present text message, wherein the step of generating includes;

(1) examining a sequence of characters of the present text message;

(2) determining an actual frequency of occurrence of the sequence of characters within the present text message; and

(3) matching the actual frequency of occurrence of the sequence of characters with one of a plurality of corecurrence libraries, wherein each of the plurality of corecurrence libraries corresponds to a different language;

c. selecting an appropriate one text-to-speech engine from a plurality of text-to-speech engines wherein each of the plurality of text-to-speech engines corresponds to one of a plurality of languages and the appropriate one text-to-speech engine is selected based upon the language identifier;

d. converting the present text message directly into computer-generated speech in response to the appropriate text-to-speech engine; and

e. playing the computer generated speech to a subscriber.

23. The method as claimed in claim 22, wherein the step of matching the actual frequency of occurrence includes:

a. comparing the actual frequency of occurrence of the sequence of characters within the present text message with a plurality of reference occurrence frequencies for the sequence of characters, wherein each reference occurrence frequency corresponds to one of the plurality of corecurrence libraries, and

b. determining a closest match between the actual frequency of occurrence of the sequence of characters and one of the reference occurrence frequencies.

24. The method as claimed in claim 22, wherein:

a. the step of examining the sequence of characters further comprises dividing the sequence of characters of the present text messages into a plurality of sequential character combinations;

b. the step of determining the actual frequency of occurrence further comprises determining a set of actual rates at which each of the plurality of sequential character combinations occur in the sequence of characters of the present text message; and

c. the step of matching the actual frequency of occurrence further comprises matching the set of actual rates with one of the plurality of corecurrence libraries.

25. The method as claimed in claim 24, wherein the step of matching the set of actual rates includes:

a. comparing the set of actual rates at which each of the plurality of sequential character combinations occur in the sequence of characters of the present text message with a plurality of reference occurrence frequencies for the sequence of characters, wherein each reference occurrence frequency corresponds to one of the plurality of corecurrence libraries, and

b. determining a closest match between the set of actual rates and one of the reference occurrence frequencies.