Speech synthesis device and method, electronic equipment and storage medium

Abstract

The invention discloses a speech synthesis device and method, electronic equipment and a computer storage medium, the speech synthesis device comprises a speech synthesis processing module and a speech enhancement module, and the output of the speech synthesis processing module is connected with the input of the speech enhancement module, wherein the speech synthesis processing module comprises a text encoder, a rhythm control module, a position sensitive attention module, an attention transition mechanism module, a decoder and a vocoder. According to the embodiment of the invention, the clear synthetic speech carrying the style rhythm can be obtained.

Description

technical field [0001] The present application relates to the technical field of speech synthesis, in particular to a speech synthesis device, method, electronic equipment and computer storage medium. Background technique [0002] Speech synthesis, also known as Text To Speech (TTS), refers to a technology that can convert input text into corresponding speech. Speech synthesis is one of the most important technologies in human-computer voice interaction. Traditional speech synthesis technology adopts a front-end model architecture. The implementation process requires detailed design of the front-end model and the back-end model, and the introduction of a large amount of expert knowledge (such as linguistics, acoustics, etc.) The logic is more complicated, and in terms of effect, there is a big gap between the naturalness of traditional synthetic speech and human speech. At present, end-to-end speech synthesis technology has emerged. End-to-end speech synthesis can obtain hi...

AI Technical Summary

Problems solved by technology

[0003] At present, when the speech synthesis method based on end-to-end speech synthesis technology uses the constructed and trained model to generate speech, it can only generate the speech of a specific speaker and has a poor sense of rhythm. Rhythmic styles such as fast and slow, joy, anger, sorrow, and joy are still difficult to simulate the rhythm of Chinese speech. Therefore, the mechanism of adding rhythm to Chinese speech needs to be optimized and improved; The recorded audio is used for synthetic training. However, it is more common in practice that many of the training audios come from artificially recorded audio. For example, in educational and teaching scenarios, the teaching audio recorded in the classroom will inevitably record noisy backgrounds. Noise, which affects the quality of model training and the quality of subsequent synthesized audio

Images