Back to results
US 9,870,769 B2Granted
Accent correction in speech recognition systems
AssigneeInternational Business Machines Corporation
InventorsSu Liu, Yi Liu, Cheng Xu, Shi Lei Zhang
Life of this patent
Estimating the term…
Abstract
A method comprising receiving an audio input signal comprising speech, determining an accent class corresponding to the speech, identifying an accented phone pattern within the speech, replacing the accented phone pattern with an unaccented phone pattern, and generating an unaccented output signal from the unaccented phone pattern.
Claims14
A method comprising:
- receiving a dictation of a prepared text;
- creating an acoustic model of the dictation of the prepared text;
- comparing the acoustic model of the dictation to an acoustic model of an unaccented pronunciation of the prepared text to provide one or more accented phone patterns corresponding to an accent class;
- updating an accent database entry corresponding to the accent class to include the one or more accented phone patterns;
- receiving an audio input signal comprising speech;
- determining an accent class corresponding to the speech;
- identifying an accented phone pattern within the speech;
- replacing the accented phone pattern with an unaccented phone pattern according to the accent database entry;
- generating an unaccented output signal from the unaccented phone pattern;
- and
- outputting the generated unaccented output signal.
8 more elements
A computer program product comprising:
- one or more computer readable storage media and program instructions stored on the one or more computer readable storage media, the program instructions comprising instructions to:
- receive a dictation of a prepared text;
- create an acoustic model of the dictation of the prepared text;
- compare an acoustic model of the dictation to an acoustic model of an unaccented pronunciation of the prepared text to provide one or more accented phone patterns corresponding to an accent class;
- update an accent database entry corresponding to the accent class to include the one or more accented phone patterns;
- receive an audio input signal comprising speech;
- determine an accent class corresponding to the speech;
- identify an accented phone pattern within the speech;
- replace the accented phone pattern with an unaccented phone pattern according to the accent database entry;
- generate an unaccented output signal from the unaccented phone pattern;
- and
- output the generated unaccented output signal.
9 more elements
A computer system comprising:
- one or more computer processors;
- one or more computer-readable storage media;
- program instructions stored on the computer-readable storage media for execution by at least one of the one or more processors, the program instructions comprising instructions to:
- receive a dictation of a prepared text;
- create an acoustic model of the dictation of the prepared text;
- compare an acoustic model of the dictation to an acoustic model of an unaccented pronunciation of the prepared text to provide one or more accented phone patterns corresponding to an accent class;
- update an accent database entry corresponding to the accent class to include the one or more accented phone patterns;
- receive an audio input signal comprising speech;
- determine an accent class corresponding to the speech;
- identify an accented phone pattern within the speech;
- replace the accented phone pattern with an unaccented phone pattern according to the accent database entry;
- generate an unaccented output signal from the unaccented phone pattern;
- and
- output the generated output signal.
11 more elements
Dependencies are inferred from the claim text.