Speech synthesis method and system based on neural network, and storage medium

Abstract

The invention discloses a speech synthesis method based on a neural network. The speech synthesis method comprises the following steps of generating a pre-trained neural network model and a specific human speech synthesis model by using a sample library; performing speech and text analysis on specific human speech text data, extracting key speech text contents in a speech text classification set,and corresponding to set labels; generating a first matching rule according to the set labels and the key speech text contents, and generating a second matching rule according to a specific person andthe specific human speech synthesis model; and calling the first matching rule and the second matching rule according to a user instruction, and outputting synthetic speech. According to the speech synthesis method based on the neural network provided by the invention, speech texts are trained, specific people can be targeted, there is less demand on the amount of data, and the speech synthesis method can be customized for use according to user needs. Moreover, for each specific human speech text data, the corresponding specific human speech synthesis model is generated, when a user input instruction contains specific person information, the corresponding model is directly called, so that the speech synthesis effect is better.

Description

technical field [0001] The present invention relates to the technical field of speech synthesis, in particular to a neural network-based speech synthesis method, system and storage medium. Background technique [0002] Speech synthesis, also known as text-to-speech (Text To Speech, TTS). Speech synthesis is the technology of producing artificial voice through mechanical and electronic methods. It is a technology that converts the text information generated by the computer itself or input from the outside into understandable and fluent spoken Chinese output. Speech synthesis is equivalent to installing a "mouth" similar to a human being on a computer, and it plays a vital role in an intelligent computer system that can "hear and speak". [0003] A traditional speech synthesis system usually includes two modules, a front-end and a back-end. The front-end analyzes the text and extracts the linguistic information required by the back-end, such as word segmentation and prosody....

AI Technical Summary

Problems solved by technology

[0004] Traditional speech synthesis systems require a large amount of sample data due to the need for parameter modeling. Not only is the acquisition cost high, but it is also difficult to obtain samples

Moreover, the traditional speech synthesis system needs to extract back-end linguistic information. For some low-quality speech, such as non-professional recording acquisition, speech with changeable emotions and spaces, speech synthesis is difficult and the synthesis effect is not ideal

Images