Voice Characterization-Based Natural Language Filtering

Abstract

An utterance is analyzed to determine a characteristic of the utterance and a transcription hypothesis is generated for the utterance. Grammar rules are then used to parse the transcription hypothesis to produce a plurality of interpretation hypotheses, each having a likelihood score. A set of authorized domains is determined based on the characteristic and the plurality of interpretation hypotheses are filtered according to the set of authorized domains. Of the remaining interpretation hypotheses, one is selected according to their likelihood scores. The characteristic may include one or more characteristics such as mood, prosody, or whether the utterance has a rising intonation.

What is claimed is: 
     
         1 . A non-transitory computer-readable medium storing a grammar rule, the grammar rule being written in a conditional grammar definition language, the rule being effective to control a grammar interpreter to interpret a transcription of a speech utterance, the grammar rule comprising an interpretation condition, the interpretation condition being dependent upon a specific identified characteristic of the speech utterance. 
     
     
         2 . The non-transitory computer-readable medium of  claim 1  wherein the interpretation condition is a Boolean enablement. 
     
     
         3 . The non-transitory computer-readable medium of  claim 1  wherein the interpretation condition determines a weighting that affects the computation of a likelihood score. 
     
     
         4 . A system comprising one or more processing devices, the one or more processing devices programmed to:
 execute a grammar interpreter enabled to interpret transcriptions of speech utterances according to grammar rules written in a conditional grammar definition language wherein the grammar rules may comprise interpretation conditions for the interpretation, the interpretation conditions being based on characteristics of the speech utterances.   
     
     
         5 . The system of  claim 4  wherein the interpretation conditions are Boolean enablements. 
     
     
         6 . The system of  claim 4  wherein the interpretation conditions determine weightings that affects the computation of likelihood scores. 
     
     
         7 . A method comprising, by a computer system comprising a processing device:
 interpreting transcriptions of speech utterances according to grammar rules written in a conditional grammar definition language, the grammar rules comprising interpretation conditions, the interpretation conditions being dependent upon specific identified characteristics of the speech utterances.   
     
     
         8 . The method of  claim 7  wherein the interpretation conditions are a Boolean enablement. 
     
     
         9 . The method of  claim 7  wherein the interpretation conditions determine weightings that affect the computation of likelihood scores. 
     
     
         10 . A method of recognizing a transcription of a speech utterance, the method comprising:
 characterizing, by a computer system, a speech utterance to determine at least one characteristic;   determining, by the computer system, from the speech utterance a plurality of phoneme sequence hypotheses;   recognizing, by the computer system, a set of transcription hypotheses from the plurality of phoneme sequence hypotheses conditioned on the at least one characteristic, each transcription hypothesis of the set of transcription hypotheses having a likelihood score associated therewith; and pruning, by the computer system, the set of transcription hypotheses to a subset of transcription hypotheses having the likelihood score thereof meeting a threshold condition   
     
     
         11 . A non-transitory computer-readable medium comprising code effective to cause one or more processors to:
 characterize a speech utterance to determine at least one characteristic;   determine from the speech utterance a plurality of phoneme sequence hypotheses;   recognize a set of transcription hypotheses from the plurality of phoneme sequence hypotheses conditioned on the at least one characteristic, each transcription hypothesis of the set of transcription hypotheses having a likelihood score associated therewith; and   prune the set of transcription hypotheses to a subset of transcription hypotheses having the likelihood score thereof meeting a threshold condition.   
     
     
         12 . The non-transitory computer-readable medium of  claim 11  wherein the at least one characteristic comprises speaker identification. 
     
     
         13 . The non-transitory computer-readable medium of  claim 11  wherein at least one characteristic comprises a classification according to one or more criteria. 
     
     
         14 . The non-transitory computer-readable medium of  claim 13  wherein the one or more criteria include one or more of age, gender, mood, and prosody. 
     
     
         15 . The non-transitory computer-readable medium of  claim 11  wherein the at least one characteristic indicates a weight. 
     
     
         16 . A non-transitory computer-readable medium comprising a grammar rule, the grammar rule being written in a conditional grammar definition language, the rule being effective to control a grammar interpreter to interpret a transcription hypothesis, the grammar rule comprising a statement that enables the interpretation conditionally when a specific characteristic of a speech utterance is identified.