Abstract: The present disclosure provides a sound-collecting method, apparatus, device and computer storage medium, wherein the method comprises: a sound-collecting apparatus collecting first sound data while playing a preset speech section; collecting sound data of a user following and reading the speech section; subjecting the sound data of following and reading the speech section to interference removal processing by using a sound interference coefficient to obtain second sound data, wherein the sound interference coefficient is determined with the speech section and the first sound data; obtaining training data for speech synthesis by using the second sound data. The quality of the collected sound data can be improved in a manner provided by the present disclosure.

Abstract: This application discloses a prosodic pause prediction method, a prosodic pause prediction device and an electronic device. The specific implementation scheme includes: obtaining a first matrix by mapping a to-be-tested text sequence through a trained embedding layer, where the to-be-tested text sequence includes a to-be-tested input text and an identity of a to-be-tested speaker; inputting the first matrix into a trained attention model, and determining a semantic representation matrix by the trained attention model; and, performing prosodic pause prediction based on the semantic representation matrix and outputting a prosodic pause prediction result of each word in the to-be-tested input text.

Abstract: This application discloses a prosodic pause prediction method, a prosodic pause prediction device and an electronic device. The specific implementation scheme includes: obtaining a first matrix by mapping a to-be-tested text sequence through a trained embedding layer, where the to-be-tested text sequence includes a to-be-tested input text and an identity of a to-be-tested speaker; inputting the first matrix into a trained attention model, and determining a semantic representation matrix by the trained attention model; and, performing prosodic pause prediction based on the semantic representation matrix and outputting a prosodic pause prediction result of each word in the to-be-tested input text.

Abstract: The present disclosure provides a sound-collecting method, apparatus, device and computer storage medium, wherein the method comprises: a sound-collecting apparatus collecting first sound data while playing a preset speech section; collecting sound data of a user following and reading the speech section; subjecting the sound data of following and reading the speech section to interference removal processing by using a sound interference coefficient to obtain second sound data, wherein the sound interference coefficient is determined with the speech section and the first sound data; obtaining training data for speech synthesis by using the second sound data. The quality of the collected sound data can be improved in a manner provided by the present disclosure.