Hybrid accelerated synthesis method of TTS system

Abstract

The invention provides a hybrid accelerated synthesis method of a TTS system. The hybrid accelerated synthesis method comprises the following steps: step 1, carrying out compliance check on a whole text requested to be synthesized; 2, judging and selecting a synthesis mode; 3, calling an existing record, and broadcasting the record; step 4, handling abnormal recording; 5, normally synthesizing a text; 6, ending a voice synthesis action; according to the method, the voice synthesis efficiency is optimized, mixed broadcasting is carried out in different synthesis modes, the calculation pressureof real-time synthesis is greatly reduced, the response speed of first sentence character broadcasting is increased, the voice broadcasting lag phenomenon is avoided, and the hardware cost of a computer is reduced.

Description

technical field [0001] The invention relates to the technical field of speech synthesis, in particular to a hybrid accelerated synthesis method of a TTS system. Background technique [0002] In recent years, with the rapid development of computer technology, speech technology has gradually integrated into people's lives. Products such as telephone robots, AI assistants, and children's story machines all rely on speech synthesis technology to broadcast clear and easy-to-understand messages in the process of human-computer interaction. machine-synthesized sounds. With the widespread application of speech technology, people have higher and higher requirements for the clarity, naturalness and fluency of synthesized voice. TTS technology is constantly innovating. The TTS synthesis method has changed from the initial splicing method to the parametric synthesis method in recent years, and then to the recently emerging end-to-end synthesis method. The effect of synthesized sound ha...

AI Technical Summary

Problems solved by technology

[0003] Due to focusing on the clarity, naturalness, fluency and other indicators of sound, the existing technology cannot eliminate the objective impact brought by the huge amount of calculation. The response speed of the first packet and the synthesis speed are getting slower and slower. The requirements for computer performance indicators are getting higher and higher

Problems such as high cost, slow response speed and easy lag directly affect the landing effect of TTS when it is applied to human-computer interaction scenarios

How to make the synthesized voice not be limited by computer performance on the basis of maintaining a high degree of anthropomorphism, improve the efficiency of TTS synthesis, and avoid the phenomenon of voice broadcast stuttering in the process of human-computer interaction has always been an important aspect of the implementation of TTS technology. problem

Images