Tencent America LLC

Microsoft Technology Licensing, LLC

Other

Chao Weng

Dong Yu

Chengzhu Yu

Jia Cui

Jun Wang

Dan Su

Heng Lu

Guangsen Wang

Peidong Wang

James G Droppo

Michael Lewis Seltzer

Kun Xu

Meng Yu

Yong Xu

Shi-Xiong Zhang

Jianming Liu

A method, computer program, and computer system is provided for converting a singing voice of a first person associated with a first speaker to a singing voice of a second person using a speaking voice of the second person associated with a second speaker. A context associated with one or more phonemes corresponding to the singing voice of a first person is encoded, and the one or more phonemes are aligned to one or more target acoustic frames based on the encoded context. One or more mel-spectrogram features are recursively generated from the aligned phonemes, the target acoustic frames, and a sample of the speaking voice of the second person. A sample corresponding to the singing voice of a first person is converted to a sample corresponding to the second singing voice using the generated mel-spectrogram features.

Learning singing from speech

Tumbler drum

Methods and apparatuses are provided for performing sequence to sequence (Seq2Seq) speech recognition training performed by at least one processor. The method includes acquiring a training set comprising a plurality of pairs of input data and target data corresponding to the input data, encoding the input data into a sequence of hidden states, performing a connectionist temporal classification (CTC) model training based on the sequence of hidden states, performing an attention model training based on the sequence of hidden states, and decoding the sequence of hidden states to generate target labels by independently performing the CTC model training and the attention model training.

Multi-task training architecture and strategy for attention-based speech recognition system

A method and apparatus are provided that analyzing sequence-to-sequence data, such as sequence-to-sequence speech data or sequence-to-sequence machine translation data for example, by minimum Bayes risk (MBR) training a sequence-to-sequence model and within introduction of applications of softmax smoothing to an N-best generation of the MBR training of the sequence-to-sequence model.

N-best softmax smoothing for minimum bayes risk training of attention based sequence-to-sequence models

A method, computer program, and computer system is provided for converting a singing first singing voice associated with a first speaker to a second singing voice associated with a second speaker. A context associated with one or more phonemes corresponding to the first singing voice is encoded, and the one or more phonemes are aligned to one or more target acoustic frames based on the encoded context. One or more mel-spectrogram features are recursively generated from the aligned phonemes and target acoustic frames, and a sample corresponding to the first singing voice is converted to a sample corresponding to the second singing voice using the generated mel-spectrogram features.

Singing voice conversion

A method of attention-based end-to-end (A-E2E) automatic speech recognition (ASR) training, includes performing cross-entropy training of a model, based on one or more input features of a speech signal, determining a posterior probability vector at a time of a first wrong token among one or more output tokens of the model of which the cross-entropy training is performed, and determining a loss of the first wrong token at the time, based on the determined posterior probability vector. The method further includes determining a total loss of a training set of the model of which the cross-entropy training is performed, based on the determined loss of the first wrong token, and updating the model of which the cross-entropy training is performed, based on the determined total loss of the training set.

Token-wise training for attention based end-to-end speech recognition

Method and apparatus for automatically predicting lexical sememes using a lexical dictionary, comprising inputting a word, retrieving the word's semantic definition and sememes corresponding to the word from an online dictionary, setting each of the retrieved sememes as a candidate sememe, inputting the word's semantic definition and candidate sememe, and estimating the probability that the candidate sememe can be inferred from the word's semantic definition.

Automatic lexical sememe prediction system using lexical dictionaries

A method, computer system, and computer readable medium are provided for automatic speech recognition. Video data and audio data corresponding to one or more speakers is received. A minimum variance distortionless response function is applied to the received audio and video data. A predicted target waveform corresponding to a target speaker from among the one or more speakers is generated based on back-propagating the output of the applied minimum variance distortionless response function.

Multi-tap minimum variance distortionless response beamformer with neural networks for target speech separation

A method, a computer readable medium, and a computer system are provided for singing voice conversion. Data corresponding to a singing voice is received. One or more features and pitch data are extracted from the received data using one or more adversarial neural networks. One or more audio samples are generated based on the extracted pitch data and the one or more features.

Growing community of inventors

Fremont, CA, United States of America

Chao Weng