Method and system for improved speech recognition by degrading utterance pronunciations

• US 7,983,914 B2
• Filed: 08/10/2005
• Issued: 07/19/2011
• Est. Priority Date: 08/10/2005
• Status: Expired due to Fees

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Ruback, Harvey M., Eckhart, John W.
• Primary Examiner(s): Dorvil; Richemond
• Assistant Examiner(s): Borsetti; Greg
• Application Number:
  US11/200,810
• Publication Number:
  US 20070038454A1
• Time in Patent Office: 2,169 Days
• Field of Search: 704/260, 704/254, 704/251, 704/243, 704/244
• US Class Current: 704/254
• CPC Class Codes: G10L 15/063 Training

Abstract

A speech recognition system or method can include a speech input device and a processor coupled to the speech input device. The processor can be programmed to identify a plurality of words that are members of confusable pairs of words where each pair includes a target word and a substituted word. The processor can degrade a pronunciation of the substituted word to provide a worse pronunciation of the substituted word. The processor can further compare the pronunciation of the target word with the worse pronunciation to the substituted word. The processor can be further programmed to reduce confusion between the substituted word and other words in a recognition grammar of the speech recognition engine and can also narrow the scope within which the substituted word is recognized.

Claims

1. In a speech recognition engine executing on a computer, a method of improving overall system recognition performance comprising steps of:

identifying a confusable pair of words, comprising a target word and a substitute word, in a candidate set of incorrectly recognized words by;

(a) identifying a first word as being a word that is rarely misrecognized and is frequently returned as confusions for other words;

(b) identifying a second word as being a word for which the first word is frequently substituted;

(c) determining the first word to be the substitute word in the confusable pair; and

(d) determining the second word to be the target word in the confusable pair; and

adjusting a pronunciation of the substitute word to a worse pronunciation of the substitute word, wherein the adjusted pronunciation of the substitute word results in less accurate detection of the substitute word by the speech recognition engine.

2. The method of claim 1, wherein the method reduces confusion between the substituted word and other words in a recognition grammar of the speech recognition engine.

3. The method of claim 1, wherein the method further comprises the step of reducing confusion between the substituted word and words in a recognition grammar other than the target word.

4. A speech recognition computer-implemented system, comprising:

a speech input device;

a processor coupled to the speech input device, wherein the processor is programmed to;

identify a confusable pair of words, comprising a target word and a substitute word, in a candidate set of incorrectly recognized words by;

(a) identifying a first word as being a word that is rarely misrecognized and is frequently returned as confusions for other words;

(b) identifying a second word as being a word for which the first word is frequently substituted;

(c) determining the first word to be the substitute word in the confusable pair; and

(d) determining the second word to be the target word in the confusable pair; and

adjust a pronunciation of the substitute word to a worse pronunciation of the substitute word, wherein the adjusted pronunciation of the substitute word results in less accurate detection of the substitute word by the speech recognition engine.

5. The speech recognition system of claim 4, wherein the processor is further programmed to reduce confusion between the substituted word and other words in a recognition grammar of the speech recognition engine.

6. The speech recognition system of claim 4, wherein the processor is further programmed to reduce confusion between the substituted word and words in a recognition grammar other than the target word.

7. A non-transitory machine-readable storage, having stored thereon a computer program having a plurality of code sections executable by a machine for causing the machine to perform the steps of:

identifying a confusable pair of words, comprising a target word and a substitute word, in a candidate set of incorrectly recognized words by;

(a) identifying a first word as being a word that is rarely misrecognized and is frequently returned as confusions for other words;

(b) identifying a second word as being a word for which the first word is frequently substituted;

(c) determining the first word to be the substitute word in the confusable pair; and

(d) determining the second word to be the target word in the confusable pair; and

adjusting a pronunciation of the substitute word to a worse pronunciation of the substitute word, wherein the adjusted pronunciation of the substitute word results in less accurate detection of the substitute word by the speech recognition engine.

8. The non-transitory machine readable storage of claim 7, wherein the computer program degrades the pronunciation by providing a worse pronunciation to the substituted word.

9. The non-transitory machine readable storage of claim 7, wherein the computer program further comprises a plurality of code sections for reducing confusion between the substituted word and words in a recognition grammar other than the target word.