Abstract: According to one embodiment, a speech recognition apparatus includes processing circuitry. The processing circuitry generates, based on sensor information, environmental information relating to an environment in which the sensor information has been acquired, generates, based on the environmental information and generic speech data, an adapted acoustic model obtained by adapting a base acoustic model to the environment, acquires speech uttered in the environment as input speech data, and subjects the input speech data to a speech recognition process using the adapted acoustic model.

Abstract: A signal processing apparatus according an embodiment includes an acquisition unit and an application unit. The acquisition unit acquires M detection signals output from M detector devices having N-fold symmetry (M is an integer equal to or greater than 2, and N is an integer equal to or greater than 2). Each of the M detector devices detects original signals generated from K signal sources (K is an integer equal to or greater than 2) having the N-fold symmetry. The application unit applies a trained neural network to M input vectors corresponding to the M detection signals and outputs K output vectors. The same parameter is set to, of multiple weights included in a weight matrix of the trained neural network, weights that are commutative based on the N-fold symmetry.

Abstract: According to one embodiment, a threshold generation method includes generating a threshold to be set in a keyword detection device. The keyword detection device detects, based on a result of comparison of the threshold with a keyword score representing a degree of similarity between voice included in an audio signal and a preset keyword, whether the audio signal includes the keyword. The threshold generation method includes: calculating keyword scores representing degrees of similarity between the keyword and a plurality of reference audio signals; calculating parameters representing a distribution of a score set including the keyword scores calculated based on the reference audio signals; and generating the threshold based on the parameters representing the distribution of the score set.

Abstract: According to one embodiment, a speech recognition apparatus includes processing circuitry. The processing circuitry generates a plurality of augmented speech data, based on input speech data, generates a plurality of acoustic scores, based on the plurality of augmented speech data and an acoustic model, generates a plurality of adjusted acoustic scores by resampling the acoustic scores, generates an integrated acoustic score by integrating the adjusted acoustic scores, generates an integrated lattice, based on the integrated acoustic score, a pronunciation dictionary, and a language model, and searches a speech recognition result with a highest likelihood from the integrated lattice.

Abstract: According to one embodiment, a moving body operation support system includes an acquirer, a microphone, a transceiver, and a processor. The acquirer acquires moving body information relating to a state of a moving body. The microphone is provided in the moving body. The transceiver performs transmitting to and receiving from an operator communication device. The processor implements one of a first operation or a second operation based on the moving body information and instruction information. The instruction information relates to an instruction based on a sound acquired by the microphone. The instruction is of a user riding in the moving body. In the first operation, the processor causes the moving body to perform an operation corresponding to the instruction information. In the second operation, the processor enables communication between the user and the operator by the transmitting and receiving between the transceiver and the operator communication device.

Abstract: According to one embodiment, a signal processing apparatus includes the following units. The transform unit transforms a first signal into a time-frequency domain to obtain a second signal, the first signal obtained by detecting sound at each of different positions. The first calculation unit calculates a first spatial correlation matrix based on the second signal. The second calculation unit calculates a second spatial correlation matrix based on a third signal obtained by delaying the second signal by a predetermined time. The spatial filter unit generates a spatial filter based on the first spatial correlation matrix and the second spatial correlation matrix, and filters the second signal by using the spatial filter.

Abstract: According to one embodiment, a dictionary editing apparatus includes processing circuitry. The processing circuitry is configured to extract words from text data, append character pronunciations to the extracted words, and specify, when a modification is made to word information including the extracted words and the appended character pronunciations, a modification candidate that is a word or character pronunciation to be modified in relation to the modification.

Abstract: According to one embodiment, a difference extraction device includes processing circuitry. The processing circuitry acquires a text in which an input notation string is described. The processing circuitry converts the input notation string into a pronunciation string. The processing circuitry executes a pronunciation string conversion process in which the pronunciation string is converted into an output notation string. The processing circuitry extracts a difference by comparing the input notation string and the output notation string with each other.

Abstract: According to one embodiment, a dictionary editing apparatus includes a processor with hardware. The processor extracts a word from text data. The processor appends a character pronunciation to the extracted word. The processor calculates at least one of a first reliability denoting a reliability of the extracted word and a second reliability denoting a reliability of the appended character pronunciation. The processor specifies a word to be a modification candidate in accordance with the first reliability. The processor specifies a character pronunciation to be a modification candidate in accordance with the second reliability.

Abstract: According to one embodiment, a signal generation device includes one or more processors. The processors convert an acoustic signal and output amplitude and phase at a plurality of frequencies. The processors, for each of a plurality of nodes of a hidden layer included in a neural network that treats the amplitude and the phase as input, obtain frequency based on a plurality of weights used in arithmetic operation of the node. The processors generate an acoustic signal based on the plurality of obtained frequencies and based on amplitude and phase corresponding to each of the plurality of nodes.

Abstract: A signal processing apparatus according an embodiment includes an acquisition unit and an application unit. The acquisition unit acquires M detection signals output from M detector devices having N-fold symmetry (M is an integer equal to or greater than 2, and N is an integer equal to or greater than 2). Each of the M detector devices detects original signals generated from K signal sources (K is an integer equal to or greater than 2) having the N-fold symmetry. The application unit applies a trained neural network to M input vectors corresponding to the M detection signals and outputs K output vectors. The same parameter is set to, of multiple weights included in a weight matrix of the trained neural network, weights that are commutative based on the N-fold symmetry.

Abstract: According to one embodiment, a speech recognition apparatus includes processing circuitry. The processing circuitry generates, based on sensor information, environmental information relating to an environment in which the sensor information has been acquired, generates, based on the environmental information and generic speech data, an adapted acoustic model obtained by adapting a base acoustic model to the environment, acquires speech uttered in the environment as input speech data, and subjects the input speech data to a speech recognition process using the adapted acoustic model.

Abstract: According to one embodiment, an information processing apparatus includes one or more processors configured to detect a trigger from a voice signal, the trigger indicating start of voice recognition; and to perform voice recognition of a recognition sound section subsequent to a trigger sound section including the detected trigger, referring to a trigger and voice recognition dictionary corresponding to the trigger.

Abstract: According to one embodiment, a sound processing apparatus extracts a feature of first speech uttered outside an objective area from first speech obtained at positions different from each other in a space of the objective area and a place outside the objective area. The apparatus creates, by learning, a determination model configured to determine whether an utterance position of second speech in the space is outside the objective area based at least in part on the feature uttered outside the objective area. The apparatus eliminates a portion of the second speech uttered outside the objective area from the second speech obtained by a second microphone based at least in part on the feature and the model. The apparatus detects and outputs remaining speech from the second speech.

Abstract: A signal processing apparatus includes one or more processors. The processors acquire a plurality of observed signals acquired from a plurality of microphone groups each including at least one microphone selected from a plurality of microphones. The microphone groups include respective microphone combinations each including at least one microphone, the combinations are different from each other, and at least one of the microphone groups includes a plurality of microphones. The processors estimate a mask indicating occupancy for each of time frequency points of a sound signal of a space corresponding to the observed signal in a plurality of spaces, for each of the observed signals. The processors integrate masks estimated for the observed signals to generate an integrated mask indicating occupancy for each of time frequency points of a sound signal in a space determined based on the spaces.

Abstract: According to one embodiment, a signal processing apparatus includes the following units. The transform unit transforms a first signal into a time-frequency domain to obtain a second signal, the first signal obtained by detecting sound at each of different positions. The first calculation unit calculates a first spatial correlation matrix based on the second signal. The second calculation unit calculates a second spatial correlation matrix based on a third signal obtained by delaying the second signal by a predetermined time. The spatial filter unit generates a spatial filter based on the first spatial correlation matrix and the second spatial correlation matrix, and filters the second signal by using the spatial filter.

Abstract: A signal processing apparatus includes one or more processors. The processors acquire a plurality of observed signals acquired from a plurality of microphone groups each including at least one microphone selected from a plurality of microphones. The microphone groups include respective microphone combinations each including at least one microphone, the combinations are different from each other, and at least one of the microphone groups includes a plurality of microphones. The processors estimate a mask indicating occupancy for each of time frequency points of a sound signal of a space corresponding to the observed signal in a plurality of spaces, for each of the observed signals. The processors integrate masks estimated for the observed signals to generate an integrated mask indicating occupancy for each of time frequency points of a sound signal in a space determined based on the spaces.

Abstract: In a speech recognition device according to one embodiment, a microphone detects sound and generates an audio signal corresponding to the sound, an adjustment processor adjusts a threshold to be a value less than a first volume level of first input audio signal generated by the microphone, and registers the adjusted threshold, a recognition processor reads the registered threshold, compares the registered threshold with a second input audio signal, discards the second input audio signal when a second volume level of the second input audio signal is less than the registered threshold, and performs a recognition process as the audio signal of a user to be recognized when the second volume level of the second input audio signal is greater than or equal to the registered threshold.

Abstract: According to one embodiment, a signal generation device includes one or more processors. The processors convert an acoustic signal and output amplitude and phase at a plurality of frequencies. The processors, for each of a plurality of nodes of a hidden layer included in a neural network that treats the amplitude and the phase as input, obtain frequency based on a plurality of weights used in arithmetic operation of the node. The processors generate an acoustic signal based on the plurality of obtained frequencies and based on amplitude and phase corresponding to each of the plurality of nodes.

Abstract: According to one embodiment, an information processing apparatus includes one or more processors configured to detect a trigger from a voice signal, the trigger indicating start of voice recognition; and to perform voice recognition of a recognition sound section subsequent to a trigger sound section including the detected trigger, referring to a trigger and voice recognition dictionary corresponding to the trigger.