Method and device for generating speech, storage medium, and electronic device
AssigneeNetease (Hangzhou) Network Co., Ltd.
InventorsHaitong Zhang, Yue Lin
Life of this patent
Estimating the term…
Abstract
A method for generating a speech includes acquiring a voice feature vector of a voice to be processed, and inputting the voice feature vector into a voice generation model to obtain a language unit vector; acquiring a text feature vector, and determining, according to the text feature vector and the language unit vector, a feature vector to be processed; and inputting the feature vector to be processed into a sequence-to-sequence model to obtain an acoustic feature vector, and inputting the acoustic feature vector into a vocoder to obtain a target voice corresponding to the voice to be processed or the text feature vector.
Claims20
A method for generating a speech, performed by terminal device, comprising:
- acquiring a speech feature vector of a to-be-processed speech, and inputting the speech feature vector into a speech generation model to obtain a language unit vector;
- acquiring a text feature vector, and determining a to-be-processed feature vector according to the text feature vector and the language unit vector;
- and
- inputting the to-be-processed feature vector into a sequence-to-sequence model to obtain an acoustic feature vector, and inputting the acoustic feature vector into a vocoder to obtain a target speech corresponding to the to-be-processed speech or the text feature vector.
A non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the following operations:
- acquiring a speech feature vector of a to-be-processed speech, and inputting the speech feature vector into a speech generation model to obtain a language unit vector:
- acquiring a text feature vector, and determining a to-be-processed feature vector according to the text feature vector and the language unit vector;
- and
- inputting the to-be-processed feature vector into a sequence-to-sequence model to obtain an acoustic feature vector, and inputting the acoustic feature vector into a vocoder to obtain a target speech corresponding to the to-be-processed speech or the text feature vector.
An electronic device, comprising:
- a processor;
- and
- a memory configured to store executable instructions of the processor;
- wherein the processor is configured to execute the following operations via executing the executable instructions:
- acquiring a speech feature vector of a to-be-processed speech, and inputting the speech feature vector into a speech generation model to obtain a language unit vector;
- acquiring a text feature vector, and determining a to-be-processed feature vector according to the text feature vector and the language unit vector;
- and
- inputting the to-be-processed feature vector into a sequence-to-sequence model to obtain an acoustic feature vector, and inputting the acoustic feature vector into a vocoder to obtain a target speech corresponding to the to-be-processed speech or the text feature vector.
Dependencies are inferred from the claim text.