Abstract: A pre-training method executed by a training apparatus includes converting an input acoustic feature amount sequence into a corresponding intermediate acoustic feature amount sequence having a first length using a first conversion model to which a conversion model parameter is provided, converting a correct answer symbol sequence to generate a first frame unit symbol sequence having the first length and generating a second frame unit symbol sequence having the first length by delaying the first frame unit symbol sequence by one frame, converting the second frame unit symbol sequence into an intermediate character feature amount sequence having the first length using a second conversion model to which a character feature amount estimation model parameter is provided, and performing label estimation using an estimation model to which an estimation model parameter is provided based on the intermediate acoustic feature amount sequence and the intermediate character feature amount sequence.

Abstract: A score integration unit 7 obtains a new score Score (l1:nb, c) that integrates a score Score (l1:nb, c) and a score Score (w1:ob, c). This new score Score (l1:nb, c) becomes a score Score (l1:nb) in a hypothesis selection unit 8. Thus, the score Score (l1:nb) can be said to take into account the score Score (w1:ob, c). In a speech recognition apparatus, first information is extracted on the basis of the score Score (l1:nb) taking into account the score Score (w1:ob, c). Thus, speech recognition with higher performance than that in the related art can be achieved.

Abstract: A probability matrix P is obtained on the basis of an acoustic feature amount sequence, the probability matrix P being the sum for all symbols cn of the product of an output probability distribution vector zn having an element corresponding to the appearance probability of each entry k of the n-th symbol cn for the acoustic feature amount sequence and an attention weight vector ?n having an element corresponding to an attention weight representing the degree of relevance of each frame t of the acoustic feature amount sequence with respect to a timing at which the symbol cn appears; a label sequence corresponding to the acoustic feature amount sequence in a case where a model parameter is provided is obtained; a CTC loss of the label sequence for a symbol sequence corresponding to the acoustic feature amount sequence is obtained using the symbol sequence and the label sequence; a KLD loss of the label sequence for a matrix corresponding to the probability matrix P is obtained using the matrix corresponding to

Abstract: Recognition results are acquired with high responsiveness without being affected by a network communication state. A speech recognition control device (1) acquires recognition results from a speech recognition device (2) with which it communicates through a network (3) and a speech recognition unit (13). A communication state measuring unit (11) measures a communication state of the network (3). A speech recognition requesting unit (12) transmits a request for a speech recognition process to each of the speech recognition device (2) and the speech recognition unit (13) with a timeout time set in accordance with an immediately prior communication state of the network (3). A recognition result output unit (14) outputs a recognition result based on a recognition result received from one or recognition results received from both of the speech recognition device (2) and the speech recognition unit (13).

Abstract: A model training device includes: a feature amount extraction unit 2 configured to extract a feature amount that corresponds to each of segments into which a first information sequence is divided by a predetermined unit; a second model calculation unit 3 configured to calculate an output probability distribution of second information when the extracted feature amounts are input to a second model; and a model update unit 4 configured to perform at least one of update of the first model based on the output probability distribution of first information calculated by the first model calculation unit and a correct unit number that corresponds to the acoustic feature amounts, and update of the second model based on the output probability distribution of second information calculated by the second model calculation unit and a correct unit number that corresponds to the first information sequence.

Abstract: Provided is a technology of learning an acoustic model with a certain degree of accuracy of sound recognition within a short calculation period.

Abstract: An acoustic model learning apparatus includes a parameter updating part configured to update a parameter of a second acoustic model on the basis of a first loss for a feature amount for training, based on output probability distribution of the second acoustic model which is a neural network acoustic model to be trained, and a second loss for a feature amount for training, based on an intermediate feature amount of a first acoustic model which is a trained neural network acoustic model and an intermediate feature amount of the second acoustic model.

Abstract: A model learning technique is provided to add a word or character at lower cost than the conventional art. A model learning apparatus includes: a storage section 32 in which a neural network model for voice recognition is stored; an addition section 33 that adds a unit corresponding to a word or character to be added, to the output layer of the neural network model read from the storage section 32; a model calculation section 30 that calculates an output probability distribution of an output from the output layer when a feature amount corresponding to the word or character is input to the neural network model where the unit corresponding to the word or character is added to the output layer; and a model update section 31 that updates the parameter of the output layer of the neural network model based on a correct unit number corresponding to the feature amount and the calculated output probability distribution.

Abstract: A model learning apparatus includes: a model calculation unit that calculates output probability distribution, which is an output from an output layer obtained when each feature amount corresponding to each task j?1, . . . , J?1 is inputted into a neural network model, where a main task is a task J and sub-tasks are tasks 1, . . . , J?1; and a multi-task type model update unit that updates a parameter of the neural network model so as to minimize a value of a loss function for the each task j?1, . . . , J?1, the value being calculated based on a correct unit number, which corresponds to each feature amount corresponding to the each task j?1, . . . , J?1, and the output probability distribution, which is calculated and corresponds to the each task j?1, . . .

Abstract: There is provided a neural network learning apparatus capable of adjusting an amount of reduction in a model size. A group parameter generating part grouping model parameters of a neural network model into arbitrarily defined groups and generating group parameters indicating features of the groups, a regularization term calculating part calculating a regularization term on an assumption that distribution of the group parameters is according to distribution defined by hyper parameters which are parameters defining distribution features, and a model updating part calculating a loss function from correct labels in teacher data, output probability distribution obtained by inputting feature values corresponding to the correct labels in the teacher data to the neural network model, and a regularization term, and updating the neural network model in a manner that a value of the loss function is decreased are included.