Pioneer Corporation

Techexperts Incorporation

Soichi Toyama

Mitsuya Komamura

Hiroshi Seo

Takehiko Shioda

Keiichi Yamauchi

Hiroaki Shibasaki

Kiyoshi Yajima

Hajime Kobayashi

Katsumi Amano

Yasunori Suzuki

Hiroyuki Ishihara

Shisei Cho

Ikuo Fujita

Hideki Amano

An operator recognition device is provided that eliminates the registration of data such as HMM data having a characteristic amount for which error in recognition occurs easily when recognizing an operator, and thus reduces the possibility of errors in recognition, and has stable recognition performance. When registering HMM data that is used when performing recognition processing, a speaker recognition deviceeliminates the registration of HMM data of a password having a characteristic amount of the spoken voice component that is similar to a characteristic amount that is indicated by HMM data that is already registered, and does not allow the registration of HMM data for which it is estimated that error in recognition will occur easily during the recognition process.

Operator recognition device, operator recognition method and operator recognition program

There is provided a voice recognition device and a voice recognition method that enhance the function of noise adaptation processing in voice recognition processing and reduce the capacity of a memory being used. Acoustic models are subjected to clustering processing to calculate the centroid of each cluster and the differential vector between the centroid and each model, model composition between each kind of assumed noise model and the calculated centroid is carried out, and the centroid of each composition model and the differential vector are stored in a memory. In the actual recognition processing, the centroid optimal to the environment estimated by the utterance environmental estimation is extracted from the memory, model restoration is carried out on the extracted centroid by using the differential vector stored in the memory, and noise adaptation processing is executed on the basis of the restored model.

Speech recognition device and speech recognition method

Before executing a speech recognition, a composite acoustic model adapted to noise is generated by composition of a noise adaptive representative acoustic model generated by noise-adaptation of each representative acoustic model and difference models stored in advance in a storing section, respectively. Then, the noise and speaker adaptive acoustic model is generated by executing speaker-adaptation to the composite acoustic model with the feature vector series of uttered speech. The renewal difference model is generated by the difference between the noise and speaker adaptive acoustic model and the noise adaptive representative acoustic model, to replace the difference model stored in the storing section therewith. The speech recognition is performed by comparing the feature vector series of the uttered speech to be recognized with the composite acoustic model adapted to noise and speaker generated by the composition of the noise adaptive representative acoustic model and the renewal difference model.

Apparatus and method for speech recognition

A contents presenting system includes: an analyzing unit which collects and analyzes user's conversation to output an analysis result; a contents acquiring unit which acquires contents from a contents database based on the analysis result; and a contents presenting unit which presents the acquired contents to the user. Since the analysis result of user's conversation includes a factor representing the environment where the user is talking, by determining contents based on the analysis result, it is possible to provide the contents which is suited for the environment where the user is present.

Contents presenting system and method

A speech synthesizing method which synthesizes speech naturally is disclosed. Standardized frame power values of an n-th frame is calculated when frame power values at head and tail frames in a phoneme are standardized. An average value of the power values sampled from the power frequency characteristics in the n-th frame at a predetermined frequency interval is set as a mean frame power value. A sum of squares of signal levels in one frame of a frequency signal from a sound source is calculated as a frame power correction value. A speech envelope signal is calculated as a function having variables of the standardized frame power values, the frame power correction value and the mean frame power value. The speech envelope signal adjusts the amplitude level of a speech waveform signal supplied from a vocal tract filter according to the level of the speech envelope signal.

Speech synthesis method

At the time of the speaker adaptation, first feature vector generation sections () generate a feature vector series [C] from which the additive noise and multiplicative noise are removed. A second feature vector generation section () generates a feature vector series [S] including the features of the additive noise and multiplicative noise. A path search section () conducts a path search by comparing the feature vector series [C] to the standard vector [a] of the standard voice HMM (). When the speaker adaptation section () conducts correlation operation on an average feature vector [S^] of the standard vector [a] corresponding to the path search result Dv and the feature vector series [S], the adaptive vector [x] is generated. The adaptive vector [x] updates the feature vector of the speaker adaptive acoustic model () used for the speech recognition.

Speech recognition system with an adaptive acoustic model

An initial combination HMMis generated from a voice HMMhaving multiplicative distortions and an initial noise HMM of additive noise, and at the same time, a Jacobian matrix J is calculated by a Jacobian matrix calculating sectionNoise variation Namh (cep), in which an estimated value Ha^(cep) of the multiplicative distortions that are obtained from voice that is actually uttered, additive noise Na(cep) that is obtained in a non-utterance period, and additive noise Nm(cep) of the initial noise HMMare combined, is multiplied by a Jacobian matrix, wherein the result of the multiplication and initial combination HMMare combined, and an adaptive HMMis generated. Thereby, an adaptive HMMthat is matched to the observation value series RNah(cep) generated from actual utterance voice can be generated in advance. When performing voice recognition by collating the observation value series RNah(cep) with adaptive HMMinfluences due to the multiplicative distortions and additive distortions are counterbalanced, wherein an effect that is equivalent to a case where voice recognition is carried out with clean voice can be obtained, and a robust voice recognition system can be achieved.

Voice recognition system

A multiplicative distortion Hm(cep) is subtracted from a voice HMM, a multiplicative distortion Ha(cep) of the uttered voice is subtracted from a noise HMMformed by HMM, and the subtraction results Sm(cep) and {Nm(cep)&#x2212;Ha (cep)} are combined with each other to thereby form a combined HMMin the cepstrum domain. A cepstrum R^a(cep) obtained by subtracting the multiplicative distortion Ha (cep) from the cepstrum Ra (cep) of the uttered voice is compared with the distribution R^m(cep) of the combined HMMin the cepstrum domain, and the combined HMM with the maximum likelihood is output as the voice recognition result.

A quantization error correcting device corrects quantization error included in audio information at the time of decoding. The audio information is divided into a plurality of frequency bands and compressive-encoded for each frequency band with bit allocation determined based on audible frequency characteristic. The device includes: a detecting unit for detecting, based on bit allocation information indicating bit allocation and encoded values of the compressive-encoded audio information, a range of quantization error indicating a range in which audio information value before compressive-encoding corresponding to the encoded value exists; and an outputting unit for outputting a decoded value corresponding to one of the encoded values based on the detected range of quantization error and the ranges of quantization errors of other correlated ones of the encoded values.

Growing community of inventors

Tsurugashima, Japan

Soichi Toyama