Speech synthetic method based on rhythm character

Abstract

The invention provides a speech synthesis method based on rhythm character, which can improve rhythm control method, and further raise Chinese speech synthesis naturalness. The speech synthesis method includes the following computer realizable steps: text processing program formed by text standardizing step, rhythm structure analysis step and language treatment step; synthetic element selecting program formed by element confirming step, matching step, pasting-up step, optimizing and screening step; speech synthesis processing program formed by base frequency outline generating step of phrase unit, base frequency outline generating step of syllable unit and intonation superposing step. According to Chinese voice character, Chinese tone and feature, and Chinese intonation and mode, the invention constructs a set of complete speech synthesis method based on rhythm character, wherein the steps and modules are all computer program treatment process, and have good universality and transportability, wide application scope and occasion.

Description

(1) Technical field [0001] The invention relates to the technical field of speech signal processing, in particular to a speech synthesis method based on prosodic features in the speech synthesis technology. (2) Background technology [0002] The existing Chinese speech synthesis method is a word-to-sound conversion using a character as a segmentation unit, or a phrase-based text-to-speech conversion using a grammatical word as a segmentation unit. In fact, when people speak, they do not use words or grammatical words as the segmentation unit, but use prosodic words as the segmentation unit. The first two speech synthesis methods will lead to relatively low naturalness of the synthesized speech output by the computer and the speech synthesis device, and the "machine smell" is too strong. This technology has entered the market on a large scale. The reason is that speech synthesis and its prosody control have the following problems: ① The naturalness of continuous synthetic s...

AI Technical Summary

Problems solved by technology

The first two speech synthesis methods will lead to relatively low naturalness of the synthesized speech output by the computer and the speech synthesis device, and the "machine smell" is too strong. This technology enters the market on a large scale

The reason is that speech synthesis and its prosody control have the following problems: ① The naturalness of continuous synthetic speech needs to be further improved; ② The text analysis process should be able to reflect the rhythm changes in natural speech to enrich the expressiveness of synthetic speech; ③ The prosody control process of speech synthesis should conform to the prosody of natural speech

In practice, it is found that the prosodic model and prosodic control of Chinese are extremely complex, and the primitive selection method based on one method cannot meet the requirements of speech synthesis for prosodic features and prosodic control in the current situation.

Images