Signal processing apparatus capable of learning a voice command which is unsuccessfully recognized and method of recognizing a voice command thereof

• US 8,812,317 B2
• Filed: 09/02/2009
• Issued: 08/19/2014
• Est. Priority Date: 01/14/2009
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: Jang, Jong-hyuk, Park, Seung-kwon, Lea, Jong-ho
• Primary Examiner(s): He, Jialong
• Application Number:
  US12/552,540
• Publication Number:
  US 20100179812A1
• Time in Patent Office: 1,812 Days
• Field of Search: 704/275, 704/270, 704/244
• US Class Current: 704/244
• CPC Class Codes: G10L 15/065 Adaptation
• CPC Class Codes: G10L 2015/0631 Creating reference templates; Clustering

Abstract

Provided are an apparatus and method for recognizing voice commands, the apparatus including: a voice command recognition unit which recognizes an input voice command; a voice command recognition learning unit which learns a recognition-targeted voice command; and a controller which controls the voice command recognition unit to recognize the recognition-targeted voice command from an input voice command, controls the voice command recognition learning unit to learn the input voice command if the voice command recognition is unsuccessful, and performs a particular operation corresponding to the recognized voice command if the voice command recognition is successful.

Claims

1. A signal processing apparatus comprising:

a voice command recognition unit which recognizes an input voice command;

a voice command recognition learning unit which learns a recognition-targeted voice command by comparing feature values of the input voice command with feature values of the recognition-targeted voice command to determine a similarity; and

a controller, which controls the voice command recognition unit to recognize the input voice command based on reference data corresponding to the recognition-targeted voice command, controls the voice command recognition learning unit to learn the input voice command based on whether the voice command recognition is successful or unsuccessful,

wherein if the voice command recognition is successful, performs an operation according to the successfully recognized voice command,

herein if the voice command recognition is unsuccessful with regard to a first input voice command, the controller determines the first input voice command as the recognition-targeted voice command according to a user input, and next time, controls the voice command recognition unit to recognize a second input voice command based on the reference data and the first input voice command, and

wherein if the voice command feature values in an unsuccessful attempt are similar to voice command feature values in a successful attempt, the voice command recognition learning unit learns the voice command feature values in the unsuccessful attempt using an Acoustic model, and if voice command feature values in the unsuccessful attempt are not similar to voice command feature values in the successful attempt, the voice command recognition learning unit adds a word having the voice command feature values in the unsuccessful attempt to a voice command recognition object using a Lexicon model.

2. The signal processing apparatus according to claim 1, wherein the controller controls the voice command recognition learning unit to learn features of the input voice command if the similarity between the input voice command and the recognition-targeted voice command is above a predetermined level.

3. The signal processing apparatus according to claim 2, wherein the voice command recognition learning unit learns the features of the input voice command using an Acoustic model.

4. The apparatus according to claim 3, wherein the Acoustic model comprises:

constructing tri-phonemes based on Acoustic model learning materials;

matching the constructed tri-phonemes with a phoneme value of the input voice command; and

changing the constructed tri-phoneme value.

5. The signal processing apparatus according to claim 1, wherein the controller controls the voice command recognition learning unit to add the input voice command to the recognition-targeted voice command if a similarity between the input voice command and the recognition-targeted voice command is below a predetermined level.

6. The signal processing apparatus according to claim 5, wherein the voice command recognition learning unit adds the input voice command to the recognition-targeted voice command using a Lexicon model.

7. The apparatus according to claim 6, wherein the Lexicon model comprises:

gather a plurality of tuples which includes final recognition words from a Lexicon model learning data;

calculate a similarity between the voice command features values of the plurality of tuples and the current input voice command feature levels;

determine if the calculated similarity is above a predetermined level;

if the similarity is above the predetermined level, add the input voice command to the voice recognition unit, and delete the corresponding tuple; and

if the similarity is below the predetermined level, add the input voice command word to the voice recognition unit.

8. The signal processing apparatus according to claim 1, further comprising a user input unit which receives information on whether the voice command recognition is unsuccessful or successful from a user.

9. A method of recognizing a voice command in a signal processing apparatus, the method comprising:

inputting a voice command;

recognizing the input voice command based on reference data corresponding to a recognition-targeted voice command;

determining a similarity between the input voice command and the recognition-targeted voice command by comparing feature values of the input voice command with feature values of the recognition-targeted voice command; and

learning the input voice command if the voice command recognition is unsuccessful and performing a particular operation corresponding to the recognized voice command if the voice command recognition is successful,

wherein if the voice command recognition is unsuccessful with regard to a first input voice command, the controller determines the first input voice command as the recognition-targeted voice command according to a user input, and next time, controls the voice command recognition unit to recognize a second input voice command based on the reference data and the first input voice command, and

wherein if the voice command feature values in an unsuccessful attempt are similar to voice command feature values in a successful attempt, learning the voice command feature values in the unsuccessful attempt using an Acoustic model, and if voice command feature values in the unsuccessful attempt are not similar to voice command feature values in the successful attempt, adding a word having the voice command feature values in the unsuccessful attempt to a voice command recognition object using a Lexicon model.

10. The method according to claim 9, wherein features of the input voice command are learned if a similarity between the input voice command and the recognition-targeted voice command is above a predetermined level.

11. The method according to claim 10, wherein the features of the input voice command are learned using an Acoustic model.

12. The method according to claim 9, wherein the input voice command is added to the recognition-targeted voice command if similarity between the input voice command and the recognition-targeted voice command is below a predetermined level.

13. The method according to claim 12, wherein the input voice command is added to the recognition-targeted voice command using a Lexicon model.

14. The method according to claim 9, further comprising receiving information on whether the voice command recognition is unsuccessful or successful from a user.

15. A signal processing apparatus comprising:

a voice command recognition unit which performs voice command recognition on an input voice command;

a voice command recognition learning unit which learns the input voice command by comparing feature values of the input voice command with feature values of the recognition-targeted voice command to determine a similarity when the input voice command is unsuccessfully recognized by the voice command recognition unit; and

a controller which performs a corresponding operation when the input voice command is successfully recognized by the voice command recognition unit, and controls the voice command recognition learning unit to learn the input voice command when the input voice command is unsuccessfully recognized by the voice command recognition unit,

wherein if the voice command recognition is unsuccessful with regard to a first input voice command, the controller determines the first input voice command as the recognition-targeted voice command according to a user input, and next time, controls the voice command recognition unit to recognize a second input voice command based on the reference data and the first input voice command, and

wherein if the voice command feature values in an unsuccessful attempt are similar to voice command feature values in a successful attempt, the voice command recognition learning unit learns the voice command feature values in the unsuccessful attempt using an Acoustic model, and if voice command feature values in the unsuccessful attempt are not similar to voice command feature values in the successful attempt, the voice command recognition learning unit adds a word having the voice command feature values in the unsuccessful attempt to a voice command recognition object using a Lexicon model.

16. A method of recognizing voice command in a signal processing apparatus, comprising:

inputting a voice command;

determining a result as to whether the input voice command is successfully or not successfully recognized;

performing a corresponding action when the input voice command is successfully recognized;

storing the determined result whether or not the input voice command is successfully recognized; and

learning the input voice command when the input voice command is not successfully recognized by comparing feature values of the input voice command with feature values of the recognition-targeted voice command to determine a similarity,

wherein if the voice command recognition is unsuccessful with regard to a first input voice command, the controller determines the first input voice command as the recognition-targeted voice command according to a user input, and next time, controls the voice command recognition unit to recognize a second input voice command based on the reference data and the first input voice command, and

wherein if the voice command feature values in an unsuccessful attempt are similar to voice command feature values in a successful attempt, learning the voice command feature values in the unsuccessful attempt using an Acoustic model, and if voice command feature values in the unsuccessful attempt are not similar to voice command feature values in the successful attempt, adding a word having the voice command feature values in the unsuccessful attempt to a voice command recognition object using a Lexicon model.