Polyphone disambiguation method, electronic equipment and computer readable storage medium

Abstract

The invention discloses a polyphone disambiguation method, electronic equipment and a computer readable storage medium, and the method comprises the steps of obtaining a target text, determining a target polyphone in the target text, and extracting a target semantic feature of the target polyphone in the target text; obtaining a plurality of matched texts containing the target polyphone, and extracting matched semantic features of the target polyphone in the matched texts, wherein the matched text at least comprises texts corresponding to target polyphones with different pronunciations; inputting the target semantic feature and the plurality of matched semantic features into a trained time sequence neural network to obtain a target hidden layer representation corresponding to the target semantic feature and a plurality of matched hidden layer representations corresponding to the plurality of matched semantic features; and determining the pronunciation of the target polyphone in the target text based on the similarity between the target hidden layer representation and the plurality of matched hidden layer representations. According to the polyphone disambiguation method provided by the invention, polyphone disambiguation suitable for small samples is realized.

Description

technical field [0001] The present application relates to the technical field of speech synthesis, and more specifically, relates to a polyphone disambiguation method, an electronic device and a computer-readable storage medium. Background technique [0002] As we all know, there are a certain number of polyphonic characters in modern Chinese, and they will have different pronunciations in different contexts. In order to accurately mark the pronunciation of Chinese characters, it is necessary to disambiguate polyphonic characters in the process of speech synthesis. [0003] In related technologies, disambiguation of polyphonic characters is performed based on statistical machine learning, and disambiguation of polyphonic characters is regarded as a classification problem in machine learning. First, a large amount of text corpus containing polyphonic phonetic annotations is collected, and then machine learning algorithms are used to extract polyphonic characters. context inf...

AI Technical Summary

Problems solved by technology

This method requires a large amount of polyphonic annotation text, but the public data available in the industry is very scarce, and the cost of collecting a considerable scale of text corpus is also very expensive

Images