An apparatus and method for recognizing paraphrases of uttered phrases, such as place names. At least one keyword contained in a speech utterance is recognized. Then, the keyword(s) contained in the speech utterance are re-recognized using a phrase including the keyword(s). Based on both recognition results, it is determined whether a paraphrase could have been uttered. If a paraphrase could have been uttered, a phrase corresponding to the paraphrase is determined as a result of speech recognition of the speech utterance.

Speech recognition apparatus and method

An apparatus includes a device for generating plural recognition candidates from a user's utterance including a series of words and a device for screening the candidates to select a few appropriate series of words from the candidates. In selecting the appropriate series of words, various factors are taken into consideration. Those factors include evaluation of sentence structures, grammar, relation among words and dialogue topics, and situations surrounding the user. For example, a dialogue topic is first determined, and then a recognition dictionary that only includes words relating to the determined topic is dynamically restructured in the system. Words included in the series of words are recognized and determined with reference to the restructured dictionary. Thus, recognition errors are minimized and determination of a correct series of words is effectively performed.

Apparatus for determining appropriate series of words carrying information to be recognized

PROBLEM TO BE SOLVED: To provide "a multilingual voice recognition device and a multilingual voice recognition dictionary creation method" capable of recognizing by original reading of a text, even when an object language for voice recognition is set to one language. SOLUTION: The multilingual voice recognition device in which a set language for voice recognition dictionary is set to one predetermined language, includes: an input means for inputting the text which is an object of voice recognition, from an external device or medium; a storage means in which the text is stored; a text-reading conversion means corresponding to a plurality of languages for creating a reading data by analyzing a name data of the text; a voice recognition dictionary in which the reading data of the name data of the text are stored; and a control means. The control means determines a language classification of the obtained name data, creates the reading data for the name data by the text-reading data conversion means according to the determined language classification, converts the created reading data to the reading data corresponding to the set language, and registers it in the voice recognition dictionary. COPYRIGHT: (C)2011,JPO&INPIT

Device for multilingual voice recognition, multilingual voice recognition dictionary creation method

PROBLEM TO BE SOLVED: To provide an on-board device 20 capable of keeping a recognition rate in voice recognition high SOLUTION: The on-board device 20 extracts and executes a command corresponding to a reference pattern which is very similar to a feature of voice uttered by a user, from a reference pattern storage section 23 The on-board device 20 receives manual operation from the user, and executes the command corresponding to the manual operation When the manual operation from the user is received from when the command corresponding to user's voice is executed till when a predetermined time period passes, the on-board device 20 additionally registers feature information which is extracted from user's voice in a feature information storage section 26, while associated with the command corresponding to the manual operation, and creates a reference pattern of the command corresponding to the additionally added feature information again from the plurality pieces of feature information stored in a feature information storage section 26 in association with the command, and registers it in the reference pattern storage section 23 COPYRIGHT: (C)2010,JPO&INPIT

Voice recognition device and voice recognition method

Network speech recognition method in English oral language machine examination system

PROBLEM TO BE SOLVED: To solve the following problem: a CMN acoustic model having learnt from a feature level after a CMN processing to a large amount of speech data for learning and a lot of time have been required so as to create an acoustic model,. SOLUTION: The acoustic model after the CMN processing is synthesized by obtaining an approximate cepstral mean (CM) of the speech data for learning, and by subtracting the obtained CM from a mean parameter of each distribution concerning the cepstrum in the acoustic model while using a model parameter in the acoustic model created without performing the CMN processing, or using statistical information obtained in creating the acoustic model. Further, speech recognition is performed by obtaining a likelihood by collating this acoustic model after CMN processing with the feature level extracted by performing the CMN processing to a speech signal for recognition. COPYRIGHT: (C)2007,JPO&INPIT

Acoustic model creating device, method, and program, speech recognition device, method, and program, and recording medium

PURPOSE: To prevent an incorrect candidate from appearing and a correct candi date from being omitted. CONSTITUTION: The word spotting speech recognizing device is equipped with a candidate rejection part 15 which rejects a recognition candidate when the precedence time of the recognition candidate outputted by a recognition likelihood calculation part 9 to the head end position of a speech signal section decided by a speech section decision part 3 for the head end position of the recognition candidate is larger than a previously set permissible time, when the delay time to the tail end position of the speech signal section decided by a speech section decision part for the tail end position of the recognition candidate is larger than a previously set permissible time, or when the head end position of the recognition candidate is behind to the tail end position of the speech signal section decided by the speech section decision part 3 or the tail end position of the recognition candidate is precedent to the head end position of the speech signal section decided by the speech section decision part 3. COPYRIGHT: (C)1994,JPO&Japio

Word spotting speech recognizing device

PROBLEM TO BE SOLVED: To accurately recognize the voices of not only a native speaker of an object language to be recognized but also of a non-native speaker. SOLUTION: This device is provided with voice models 16 and 21 of native speakers, a voice model 23 obtained when a native speaker speaks another language, means 14, 19, 22, and 24 which calculates matching scores, based upon voice feature parameters extracted from an input voice signal with a matching pattern using the voice modles obtained from a native speaker of the object language to be recognized, a matching pattern using the voice model of the object language to be recognized by a non-native speaker, and a matching pattern generated so that the recognition object can be recognized by using other language voice models by the native speakers of other language, and a means 17 which outputs a recognized object that a matching pattern with high scores represents. COPYRIGHT: (C)2001,JPO

Method and device for voice recognition

PROBLEM TO BE SOLVED: To automatically and appropriately set strength of a deblocking filter by applying a coder, a coding method, a program of the coding method and a recording medium with the program of the coding method recorded thereon, for example, to a coder by ITU-T recommendation H. 264. SOLUTION: The present invention calculates a characteristic curve indicating the strength of the deblocking filter using a data amount Bit rate per unit time as a variable from difficulty Mad of intra coding and difficulty Sad of inter coding and calculates the strength y (α,β) of the blocking filter from the characteristic curve. COPYRIGHT: (C)2009,JPO&INPIT

明弘今村

Papers

Acoustic model creating device, method, and program, speech recognition device, method, and program, and recording medium

Word spotting speech recognizing device

Method and device for voice recognition

Coder, coding method, program of coding method and recording medium with program of coding method recorded thereon

明弘 今村

Papers

Acoustic model creating device, method, and program, speech recognition device, method, and program, and recording medium

Word spotting speech recognizing device

Method and device for voice recognition

Coder, coding method, program of coding method and recording medium with program of coding method recorded thereon

明弘今村