System and method for speech-to-text conversion using constrained dictation in a speak-and-spell mode

• US 7,676,364 B2
• Filed: 03/21/2005
• Issued: 03/09/2010
• Est. Priority Date: 03/25/2004
• Status: Active Grant

• Current Assignee: Ashwin Rao
• Original Assignee: Ashwin Rao
• Inventors: Rao, Ashwin
• Primary Examiner(s): Chawan; Vijay B
• Application Number:
  US11/084,964
• Publication Number:
  US 20050216272A1
• Time in Patent Office: 1,814 Days
• Field of Search: 704/251, 704/235, 704/270, 704/275, 704/231, 704/255, 704/E15.04, 704/E15.044, 379/88.14, 379/88.13
• US Class Current: 704/235
• CPC Class Codes: G10L 15/08 Speech classification or search
• CPC Class Codes: G10L 15/193 Formal grammars, e.g. finite state automata, context free grammars or word networks
• CPC Class Codes: G10L 2015/228 of application context

Abstract

For improving the accuracy of a speech recognition system, for the specific task of speech-to-text (dictation style speech) translation, a constrained dictation methodology using speak-and-spell mode is disclosed. The invention is perfectly suited for modern day “text-messaging” applications wherein the number of words being dictated is very small (limited by the 140-160 characters message length constraint). Additionally, the invention adds a control on the way users interact with machines, thereby making the speech recognition task easier and improving system accuracy.

Claims

1. A system for converting speech to text comprising:

a) an interface requesting a user to speak a word and to speak a spelling of said word for each of a plurality of words spoken during a communication;

b) an audio receiving module for receiving said spoken word and spoken spelling of said word;

c) a signal processing module for extracting specific features of the spoken word and said spelled word;

d) a search module that uses said extracted specific features in conjunction with an at least one acoustic module;

e) at least one language module that matches a result presented by the search module; and

f) at least one constrained lexicon that takes the resulting match and using a system module outputting the desired text matching the spoken word.

2. The system of claim 1 wherein the at least one language module incorporates a network of phonemes followed by a network of alphabets.

3. A method for converting speech to text comprising:

a) requesting a user to speak a word and to speak a spelling of said word for each of a plurality of words spoken during a communication;

b) receiving said spoken word and spoken spelling of said word using an audio receiving module;

c) extracting specific features of the spoken word and said spelling of said spoken word using a signal processing module;

d) using said extracted specific features in conjunction with an at least acoustic module using a search module;

e) matching a result presented by the search module using an at least one language module; and

f) taking the results that match an at least one constrained lexicon outputting the desired text matching the spoken word.

4. The method of claim 3 wherein the at least one language module incorporates a network of phonemes followed by a network of alphabets.

5. A means for converting speech to text comprising:

a) means for requesting a user to speak a word and to speak a spelling of said word for each of a plurality of words spoken during a communication;

b) means for receiving said spoken word and spoken spelling of said word using an audio receiving module;

c) means for extracting specific features of the spoken word and said spelling of said spoken word using a signal processing module;

d) means for using said extracted specific features in conjunction with an at least one acoustic module using a search module;

e) means for matching a result presented by the search module using an at least one language module; and

f) means for taking the results that match an at least one constrained lexicon outputting the desired text matching the spoken word.

6. The means of claim 5 wherein the at least one language module incorporates a network of phonemes means followed by a network of alphabet means.