Abstract: A noise suppression device includes processing circuitry to generate post-noise suppression data by performing a noise suppression process on input data; to determine a weighting coefficient based on the input data in a predetermined section in a time series and the post-noise suppression data in the predetermined section; and to generate output data by performing weighted addition on the input data and the post-noise suppression data by using values based on the weighting coefficient as weights.

Abstract: A speech separation device (12) of a speech separation system includes a feature amount extraction unit (121) configured to extract time-series data of a speech feature amount of mixed speech, a block division unit (122) configured to divide the time-series data of the speech feature amount into blocks having a certain time width, a speech separation neural network (1b) configured to create time-series data of a mask of each of a plurality of speakers from the time-series data of the speech feature amount divided into blocks, and a speech restoration unit (123) configured to restore the speech data of each of the plurality of speakers from the time-series data of the mask and the time-series data of the speech feature amount of the mixed speech.

Abstract: A likelihood of voice that is a discrimination measure between noise and voice is corrected, by using a Gaussian mixture model of noise learned in a time section in which an input signal is noise, and a voice activity is detected on the basis of the corrected likelihood of voice.

Abstract: An information processing device includes an acquisition unit that acquires a sound signal and a control unit that segments the sound signal into a plurality of sections, calculates a variation value as a variation amount per section time in regard to each of the plurality of sections based on the sound signal, identifies sections where the variation value is less than or equal to a predetermined threshold value among the plurality of sections, calculates power of the sound signal in each of the identified sections based on the sound signal, determines a maximum value among values of the power of the sound signal in each of the identified sections, sets a value based on the maximum value as a detection threshold value, and detects sections where the power of the sound signal with elapse of time is higher than or equal to the detection threshold value as detection target sections.

Abstract: A voice recognition device includes: a first feature vector calculating unit (2) for calculating a first feature vector from voice data input; an acoustic likelihood calculating unit (4) for calculating an acoustic likelihood of the first feature vector by using an acoustic model used for calculating an acoustic likelihood of a feature vector; a second feature vector calculating unit (3) for calculating a second feature vector from the voice data; a noise degree calculating unit (6) for calculating a noise degree of the second feature vector by using a discriminant model used for calculating a noise degree indicating whether a feature vector is noise or voice; a noise likelihood recalculating unit (8) for recalculating an acoustic likelihood of noise on the basis of the acoustic likelihood of the first feature vector and the noise degree of the second feature vector; and a collation unit (9) for performing collation with a pattern of a vocabulary word to be recognized, by using the acoustic likelihood calcula

Abstract: A speech separation device (12) of a speech separation system includes a feature amount extraction unit (121) configured to extract time-series data of a speech feature amount of mixed speech, a block division unit (122) configured to divide the time-series data of the speech feature amount into blocks having a certain time width, a speech separation neural network (1b) configured to create time-series data of a mask of each of a plurality of speakers from the time-series data of the speech feature amount divided into blocks, and a speech restoration unit (123) configured to restore the speech data of each of the plurality of speakers from the time-series data of the mask and the time-series data of the speech feature amount of the mixed speech.

Abstract: A likelihood of voice that is a discrimination measure between noise and voice is corrected, by using a Gaussian mixture model of noise learned in a time section in which an input signal is noise, and a voice activity is detected on the basis of the corrected likelihood of voice.

Abstract: A voice recognition device includes: a first feature vector calculating unit (2) for calculating a first feature vector from voice data input; an acoustic likelihood calculating unit (4) for calculating an acoustic likelihood of the first feature vector by using an acoustic model used for calculating an acoustic likelihood of a feature vector; a second feature vector calculating unit (3) for calculating a second feature vector from the voice data; a noise degree calculating unit (6) for calculating a noise degree of the second feature vector by using a discriminant model used for calculating a noise degree indicating whether a feature vector is noise or voice; a noise likelihood recalculating unit (8) for recalculating an acoustic likelihood of noise on the basis of the acoustic likelihood of the first feature vector and the noise degree of the second feature vector; and a collation unit (9) for performing collation with a pattern of a vocabulary word to be recognized, by using the acoustic likelihood calcula

Abstract: An acoustic model training device includes: a processor to execute a program; and a memory to store the program which, when executed by the processor, performs processes of: generating, based on feature vectors obtained by analyzing utterance data items of a plurality of speakers, a training data item of each speaker by subtracting, for each speaker, a mean vector of all the feature vectors of the speaker from each of the feature vectors of the speaker; generating a training data item of all the speakers by subtracting a mean vector of all the feature vectors of all the speakers from each of the feature vectors of all the speakers; and training an acoustic model using the training data item of each speaker and the training data item of all the speakers.

Abstract: An acoustic model training device includes: a processor to execute a program; and a memory to store the program which, when executed by the processor, performs processes of: generating, based on feature vectors obtained by analyzing utterance data items of a plurality of speakers, a training data item of each speaker by subtracting, for each speaker, a mean vector of all the feature vectors of the speaker from each of the feature vectors of the speaker; generating a training data item of all the speakers by subtracting a mean vector of all the feature vectors of all the speakers from each of the feature vectors of all the speakers; and training an acoustic model using the training data item of each speaker and the training data item of all the speakers.

Abstract: An apparatus includes a lip image recognition unit 103 to recognize a user state from image data which is information other than speech; a non-speech section deciding unit 104 to decide from the recognized user state whether the user is talking; a speech section detection threshold learning unit 106 to set a first speech section detection threshold (SSDT) from speech data when decided not talking, and a second SSDT from the speech data after conversion by a speech input unit when decided talking; a speech section detecting unit 107 to detect a speech section indicating talking from the speech data using the thresholds set, wherein if it cannot detect the speech section using the second SSDT, it detects the speech section using the first SSDT; and a speech recognition unit 108 to recognize speech data in the speech section detected, and to output a recognition result.

Abstract: There are provided a recognition result candidate comparator 205 that compares a plurality of server-side voice recognition result candidates received by a receiver 204, to detect texts having a difference, and a recognition result integrator 206 that integrates a client-side voice recognition result candidate and a server-side voice recognition result candidate on the basis of the client-side voice recognition result candidate, the server-side voice recognition result candidate, and a detection result provided by the recognition result candidate comparator 205, to decide a voice recognition result.

Abstract: Disclosed is a speech search device including a recognizer 2 that refers to an acoustic model and language models having different learning data and performs voice recognition on an input speech, to acquire a recognized character string for each language model, a character string comparator 6 that compares the recognized character string for each language models with the character strings of search target words stored in a character string dictionary, and calculates a character string matching score showing the degree of matching of the recognized character string with respect to each of the character strings of the search target words, to acquire both a character string having the highest character string matching score and this character string matching score for each recognized character strings, and a search result determinator 8 that refers to the acquired score and outputs one or more search target words in descending order of the scores.

Abstract: There are provided a recognition result candidate comparator 205 that compares a plurality of server-side voice recognition result candidates received by a receiver 204, to detect texts having a difference, and a recognition result integrator 206 that integrates a client-side voice recognition result candidate and a server-side voice recognition result candidate on the basis of the client-side voice recognition result candidate, the server-side voice recognition result candidate, and a detection result provided by the recognition result candidate comparator 205, to decide a voice recognition result.

Abstract: With respect to speech data 4 of an input speech 2, a speech-recognition device 1 performs at an internal recognizer 7, recognition processing using an acoustic model 9, to calculate an internal recognition result 10 and its acoustic likelihood. A reading-addition processor 12 acquires an external recognition result 11 from recognition processing of the speech data 4 of the input speech 2 by an external recognizer 19 and adds a reading thereto, and a re-collation processor 15 calculates, using the acoustic model 9, the acoustic likelihood of the external recognition result 11 to provide a re-collation result 16. A result-determination processor 17 compares the acoustic likelihood of the internal recognition result 10 with the acoustic likelihood of the external recognition result 11 included in the re-collation result 16, to thereby determine a final recognition result 18.

Abstract: With respect to speech data 4 of an input speech 2, a speech-recognition device 1 performs at an internal recognizer 7, recognition processing using an acoustic model 9, to calculate an internal recognition result 10 and its acoustic likelihood. A reading-addition processor 12 acquires an external recognition result 11 from recognition processing of the speech data 4 of the input speech 2 by an external recognizer 19 and adds a reading thereto, and a re-collation processor 15 calculates, using the acoustic model 9, the acoustic likelihood of the external recognition result 11 to provide a re-collation result 16. A result-determination processor 17 compares the acoustic likelihood of the internal recognition result 10 with the acoustic likelihood of the external recognition result 11 included in the re-collation result 16, to thereby determine a final recognition result 18.

Abstract: An index generating unit divides each name data of search target data both into words and into characters, calculates start and end scores showing a start and an end of each of the words and start and end scores showing a start and an end of each of the characters, links them to each entry word which constructs the name data as a list (a name ID, a position, and start and end scores), and stores this list in an index storage unit. A searching unit decomposes an input character string into partial character strings, acquires corresponding candidate entries from the index storage unit, and judges the continuity between candidate entries on the basis of lists to add a comparison score according to the continuity to a candidate entry.

Abstract: Voice recognition is realized by a pattern matching with a voice pattern model, and when a large number of paraphrased words are required for one facility, such as a name of a hotel or a tourist facility, the pattern matching needs to be performed with the voice pattern models of all the paraphrased words, resulting in an enormous amount of calculation. Further, it is difficult to generate all the paraphrased words, and a large amount of labor is required.

Abstract: An index generating unit divides each name data of search target data both into words and into characters, calculates start and end scores showing a start and an end of each of the words and start and end scores showing a start and an end of each of the characters, links them to each entry word which constructs the name data as a list (a name ID, a position, and start and end scores), and stores this list in an index storage unit. A searching unit decomposes an input character string into partial character strings, acquires corresponding candidate entries from the index storage unit, and judges the continuity between candidate entries on the basis of lists to add a comparison score according to the continuity to a candidate entry.

Abstract: A text data search using a voice is conventionally a full-text search using a word as an index word for a part recognized as a word in an input voice. Therefore, if any of the parts recognized as the words is falsely recognized, a search precision is lowered. In the present invention, referring to a language model generated by a language model generating part from text data to be subjected to a search which is divided by a learning data dividing part into a linguistic part and an acoustic model obtained by modeling voice features, a voice recognition part performs voice recognition for the input voice to output a phonemic representation. A matching unit converting part divides the phonemic representation into the same units as those of a text search dictionary, which is obtained by dividing the text data to be subjected to the search into the units smaller than those of the language model. A text search part uses the result of division to make a search on the text search dictionary.