Sony Corporation

Sony Electronics Inc

Other

Lex S Olorenshaw

Xavier Menendez-Pidal

Duanpei Wu

Ruxin Chen

Gustavo Hernandez Abrego

Miyuki Tanaka

Lei Duan

Michael Emonts

Jingwen Lu

Eugene Koontz

Mariscela Amador-Hernandez

Xavier Menedez-Pidal

A system and method for implementing a refined dictionary for speech recognition includes a database analyzer that initially identifies first vocabulary words that are present in a training database and second vocabulary words that are not present in the training database. A relevance module then performs refinement procedures upon the first vocabulary words to produce refined short word pronunciations and refined long word pronunciations that are added to a refined dictionary. A consensus module compares the second pronunciations with calculated plurality pronunciations to identify final consensus pronunciations that are then included in the refined dictionary.

System and method for implementing a refined dictionary for speech recognition

The present invention comprises a system and method for speech recognition utilizing a multi-language dictionary, and may include a recognizer that is configured to compare input speech data to a series of dictionary entries from the multi-language dictionary to detect a recognized phrase or command. The multi-language dictionary may be implemented with a mixed-language technique that utilizes dictionary entries which incorporate multiple different languages such as Cantonese and English. The speech recognizer may thus advantageously achieve more accurate speech recognition accuracy in an efficient and compact manner.

System and method for performing speech recognition by utilizing a multi-language dictionary

A system and method for effectively performing speech recognition procedures includes enhanced demiphone acoustic models that a speech recognition engine utilizes to perform the speech recognition procedures. The enhanced demiphone acoustic models each have three states that are collectively arranged to form a preceding demiphone and a succeeding demiphone. An acoustic model generator may utilize a decision tree for analyzing speech context information from a training database. The acoustic model generator then effectively configures each of the enhanced demiphone acoustic models as either a succeeding-dominant enhanced demiphone acoustic model or a preceding-dominant enhanced demiphone acoustic model to accurately model speech characteristics.

Methodology for generating enhanced demiphone acoustic models for speech recognition

A system and method for effectively implementing an optimized language model for speech recognition includes initial language models each created by combining source models according to selectable interpolation coefficients that define proportional relationships for combining the source models. A rescoring module iteratively utilizes the initial language models to process input development data for calculating word-error rates that each correspond to a different one of the initial language models. An optimized language model is then selected from the initial language models by identifying an optimal word-error rate from among the foregoing word-error rates. The speech recognizer may then utilize the optimized language model for effectively performing various speech recognition procedures.

System and method for effectively implementing an optimized language model for speech recognition

The present invention comprises a system and method for implementing a Mandarin Chinese speech recognizer with an optimized phone set, and may include a recognizer configured to compare input speech data to phone strings from a vocabulary dictionary that is implemented according to an optimized Mandarin Chinese phone set. The optimized Mandarin Chinese phone set may be implemented with a phonetic technique to separately include consonantal phones and vocalic phones. For reasons of system efficiency, the optimized Mandarin Chinese phone set may preferably be implemented in a compact manner to include only a minimum required number of consonantal phones and vocalic phones to accurately represent Mandarin Chinese speech during the speech recognition procedure.

System and method for Mandarin Chinese speech recognition using an optimized phone set

The present invention comprises a system and method for implementing a Cantonese speech recognizer with an optimized phone set, and may include a recognizer configured to compare input speech data to phone strings from a vocabulary dictionary that is implemented according to an optimized Cantonese phone set. The optimized Cantonese phone set may be implemented with a phonetic technique to separately include consonantal phones and vocalic phones. For reasons of system efficiency, the optimized Cantonese phone set may preferably be implemented in a compact manner to include only a minimum required number of consonantal phones and vocalic phones to accurately represent Cantonese speech during the speech recognition procedure.

System and method for cantonese speech recognition using an optimized phone set

The present invention comprises a system and method for effectively implementing a Mandarin Chinese speech recognition dictionary, and may include a recognizer configured to compare input speech data to phone strings from a vocabulary dictionary that is implemented according to an optimized Mandarin Chinese phone set. The optimized Mandarin Chinese phone set may efficiently be implemented by utilizing an allophone and phonemic variation technique. In addition, the foregoing vocabulary dictionary may be implemented by utilizing unified dictionary optimization techniques to provide robust and accurate speech recognition. Furthermore, the vocabulary dictionary may be implemented as an optimized dictionary to accurately recognize either Northern Mandarin Chinese speech or Southern Mandarin Chinese speech during the speech recognition procedure.

System and method for effectively implementing a Mandarin Chinese speech recognition dictionary

A system and method for performing a refinement procedure to effectively implement a speech recognition dictionary for spontaneous speech recognition may include a problematic word identifier configured to divide vocabulary words from an initial speech recognition dictionary into problematic words and non-problematic words according to pre-defined identification criteria. A candidate generator may analyze the problematic words to produce one or more pronunciation candidates for each of the problematic words. An optimization module may then perform an optimization process for refining one or more pronunciation candidates according to certain optimization criteria to thereby generate optimized problematic pronunciations. A dictionary refinement manager may finally combine the optimized problematic pronunciations with non-problematic pronunciations of the non-problematic words to produce a refined speech recognition dictionary for use by the speech recognition system.

Methodology for performing a refinement procedure to implement a speech recognition dictionary

A system and method for utilizing speech recognition to efficiently perform data indexing procedures includes an authoring module that coordinates an authoring procedure for creating an index file that has pattern word sets corresponding to data objects stored in a memory of a host electronic device. The pattern word sets are generated with a speech recognition engine that transforms spoken data descriptions into text data descriptions for creating the pattern word sets. The pattern word sets are associated in the index file with data object identifiers that uniquely identify the corresponding data objects. A retrieval module manages a retrieval procedure in which the speech recognition engine converts a spoken data request into a text data request. The retrieval module compares the text data request and the pattern word sets to identify a requested object identifier for locating a requested data object from among the data objects stored in the memory of the host electronic device.

System and method for utilizing speech recognition to efficiently perform data indexing procedures

The present invention comprises a system and method for speech recognition utilizing a merged dictionary, and may include a recognizer that is configured to compare input speech data to a series of dictionary entries from the merged dictionary to detect a recognized phrase or command. The merged dictionary may be implemented by utilizing a merging technique that maps two or more related phrases or commands with similar meanings to a single one of the dictionary entries. The recognizer may thus achieve more accurate speech recognition accuracy by merging phrases or commands which might otherwise be erroneously mistaken for each other.

System and method for speech recognition utilizing a merged dictionary

A system and method for speech recognition using an enhanced phone set comprises speech data, an enhanced phone set, and a transcription generated by a transcription process. The transcription process selects appropriate phones from the enhanced phone set to represent acoustic-phonetic content of the speech data. The enhanced phone set includes base-phones and composite-phones. A phone dataset includes the speech data and the transcription. The present invention also comprises a transformer that applies transformation rules to the phone dataset to produce a transformed phone dataset. The transformed phone dataset may be utilized in training a speech recognizer, such as a Hidden Markov Model. Various types of transformation rules may be applied to the phone dataset of the present invention to find an optimum transformed phone dataset for training a particular speech recognizer.

System and method for speech recognition using an enhanced phone set

The present invention comprises a methodology for implementing a vocabulary set for use in a speech recognition system, and may preferably include a recognizer for analyzing utterances from the vocabulary set to generate N-best lists of recognition candidates. The N-best lists may then be utilized to create an acoustical matrix configured to relate said utterances to top recognition candidates from said N-best lists, as well as a lexical matrix configured to relate the utterances to the top recognition candidates from the N-best lists only when second-highest recognition candidates from the N-best lists are correct recognition results. An utterance ranking may then preferably be created according to composite individual error/accuracy values for each of the utterances. The composite individual error/accuracy values may preferably be derived from both the acoustical matrix and the lexical matrix. Lowest-ranked utterances from the foregoing utterance ranking may preferably be repeatedly eliminated from the vocabulary set when a total error/accuracy value for all of the utterances fails to exceed a predetermined threshold value.

Methodology for implementing a vocabulary set for use in a speech recognition system

A system and method for speech verification using out-of-vocabulary models includes a speech recognizer that has a model bank with system vocabulary word models, a garbage model, and one or more noise models. The model bank may reject an utterance or other sound as an invalid vocabulary word when the model bank identifies the utterance or other sound as corresponding to the garbage model or the noise models. Initial noise models may be selectively combined into a pre-determined number of final noise model clusters to effectively reduce the number of noise models that are utilized by the model bank of the speech recognizer to verify system vocabulary words.

System and method for speech verification using out-of-vocabulary models

A method for utilizing validity constraints in a speech endpoint detector comprises a validity manager that may utilize a pulse width module to validate utterances that include a plurality of energy pulses during a certain time period. The validity manager also may utilize a minimum power module to ensure that speech energy below a pre-determined level is not classified as a valid utterance. In addition the validity manager may use a duration module to ensure that valid utterances fall within a specified duration. Finally, the validity manager may utilize a short-utterance minimum power module to specifically distinguish an utterance of short duration from background noise based on the energy level of the short utterance.

Method for utilizing validity constraints in a speech endpoint detector

The present invention comprises a system and method for speech verification using a confidence measure that includes a speech verifier which compares a differential score for a recognized word to a predetermined threshold value, where a recognized word is the word model that produced the highest recognition score. In one embodiment, a single threshold is used for each word in a vocabulary. In another embodiment, each word model has an associated threshold, so that a differential score for a recognized word is compared to a unique threshold associated with that word. In a further embodiment, pairs of confused words in the vocabulary are dealt with separately. If a confused word is the recognized word, the speech verifier compares the differential score to a threshold that depends on the word model that produced the next-highest recognition score. Different values for the various thresholds may maximize rejection accuracy or recognition accuracy. A trade-off between rejection accuracy and recognition accuracy may be made by utilizing an intermediate threshold value that is between a minimum threshold value and a maximum threshold value.

Growing community of inventors

Half Moon Bay, CA, United States of America

Lex S Olorenshaw