Abstract: A method for performing multi-label classification includes extracting a feature vector from an input vector including input data by a feature extractor, determining, by a label predictor, a relevant vector including relevant labels having relevant scores based on the feature vector, updating a binary masking vector by masking pre-selected labels having been selected in previous label selections, applying the updated binary masking vector to the relevant vector such that the relevant label vector is updated to exclude the pre-selected labels from the relevant labels, and selecting a relevant label from the updated relevant label vector based on the relevant scores of the updated relevant label vector.

Abstract: A method and system processes utterances that are acquired either from an automatic speech recognition (ASR) system or text. The utterances have associated identities of each party, such as role A utterances and role B utterances. The information corresponding to utterances, such as word sequence and identity, are converted to features. Each feature is received in an input layer of a neural network (NN). A dimensionality of each feature is reduced, in a projection layer of the NN, to produce a reduced dimensional feature. The reduced dimensional feature is processed to provide probabilities of labels for the utterances.

Abstract: A method and for training a language model to reduce recognition errors, wherein the language model is a recurrent neural network language model (RNNLM) by first acquiring training samples. An automatic speech recognition system (ASR) is appled to the training samples to produce recognized words and probabilites of the recognized words, and an N-best list is selected from the recognized words based on the probabilities. determining word erros using reference data for hypotheses in the N-best list. The hypotheses are rescored using the RNNLM. Then, we determine gradients for the hypotheses using the word errors and gradients for words in the hypotheses. Lastly, parameters of the RNNLM are updated using a sum of the gradients.

Abstract: Provided is a DNN learning method that can reduce DNN learning time using data belonging to a plurality of categories. The method includes the steps of training a language-independent sub-network 120 and language-dependent sub-networks 122 and 124 with training data of Japanese and English. This step includes: a first step of training a DNN obtained by connecting neurons in an output layer of the sub-network 120 with neurons in an input layer of sub-network 122 with Japanese training data; a step of forming a DNN by connecting the sub-network 124 in place of the sub-network 122 to the sub-network 120 and training it with English data; repeating these steps alternately until all training data ends; and after completion, separating the first sub-network 120 from other sub-networks and storing it as a category-independent sub-network in a storage medium.

Abstract: A method and system processes utterances that are acquired either from an automatic speech recognition (ASR) system or text. The utterances have associated identities of each party, such as role A utterances and role B utterances. The information corresponding to utterances, such as word sequence and identity, are converted to features. Each feature is received in an input layer of a neural network(NN). A dimensionality of each feature is reduced, in a projection layer of the NN, to produce a reduced dimensional feature. The reduced dimensional feature is processed to provide probabilities of labels for the utterances.

Abstract: A conventional speech recognition dictionary, translation dictionary and speech synthesis dictionary used in speech translation have inconsistencies.

Abstract: Provided is a DNN learning method that can reduce DNN learning time using data belonging to a plurality of categories. The method includes the steps of training a language-independent sub-network 120 and language-dependent sub-networks 122 and 124 with training data of Japanese and English. This step includes: a first step of training a DNN obtained by connecting neurons in an output layer of the sub-network 120 with neurons in an input layer of sub-network 122 with Japanese training data; a step of forming a DNN by connecting the sub-network 124 in place of the sub-network 122 to the sub-network 120 and training it with English data; repeating these steps alternately until all training data ends; and after completion, separating the first sub-network 120 from other sub-networks and storing it as a category-independent sub-network in a storage medium.

Abstract: Appropriate processing results or appropriate apparatuses can be selected with a control device that selects the most probable speech recognition result by using speech recognition scores received with speech recognition results from two or more speech recognition apparatuses; sends the selected speech recognition result to two or more translation apparatuses respectively; selects the most probable translation result by using translation scores received with translation results from the two or more translation apparatuses; sends the selected translation result to two or more speech synthesis apparatuses respectively; receives a speech synthesis processing result including a speech synthesis result and a speech synthesis score from each of the two or more speech synthesis apparatuses; selects the most probable speech synthesis result by using the scores; and sends the selected speech synthesis result to a second terminal apparatus.

Abstract: In conventional network-type speech translation systems, devices or models for recognizing or synthesizing speech cannot be changed in accordance with speakers' attributes, and therefore, accuracy is reduced or inappropriate output occurs in each process of speech recognition, translation, and speech synthesis. Accuracy of each processing of speech translation, translation, or speech synthesis is improved and appropriate output is performed in a network-type speech translation system by, based on speaker attributes, appropriately changing the server to perform speech recognition or the speech recognition model, appropriately changing the translation server to perform translation or the translation model, or appropriately changing the speech synthesis server or speech synthesis model.

Abstract: Appropriate processing results or appropriate apparatuses can be selected with a control device that selects the most probable speech recognition result by using speech recognition scores received with speech recognition results from two or more speech recognition apparatuses; sends the selected speech recognition result to two or more translation apparatuses respectively; selects the most probable translation result by using translation scores received with translation results from the two or more translation apparatuses; sends the selected translation result to two or more speech synthesis apparatuses respectively; receives a speech synthesis processing result including a speech synthesis result and a speech synthesis score from each of the two or more speech synthesis apparatuses; selects the most probable speech synthesis result by using the scores; and sends the selected speech synthesis result to a second terminal apparatus.

Abstract: In conventional network-type speech translation systems, devices or models for recognizing or synthesizing speech cannot be changed in accordance with speakers' attributes, and therefore, accuracy is reduced or inappropriate output occurs in each process of speech recognition, translation, and speech synthesis. Accuracy of each processing of speech translation, translation, or speech synthesis is improved and appropriate output is performed in a network-type speech translation system by, based on speaker attributes, appropriately changing the server to perform speech recognition or the speech recognition model, appropriately changing the translation server to perform translation or the translation model, or appropriately changing the speech synthesis server or speech synthesis model.

Abstract: A conventional speech recognition dictionary, translation dictionary and speech synthesis dictionary used in speech translation have inconsistencies.