Unlock instant, AI-driven research and patent intelligence for your innovation.

Polyphone pronunciation prediction method and device and computer readable storage medium

A prediction method and prediction device technology, which are applied in neural learning methods, speech analysis, speech synthesis, etc., can solve the problems of low accuracy, low accuracy of polyphonic pronunciation prediction, large corpus matching ambiguity, etc. Effect

Inactive Publication Date: 2020-08-28
NANJING SILICON INTELLIGENCE TECH CO LTD
View PDF5 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The previous pronunciation prediction methods for polyphonic characters mainly include: 1. Directly follow the pronunciation with the highest frequency, obviously, this method has a low accuracy rate; 2. Summarize the polyphonic word lexicon and corpus, and then use the phrase matching method to process polyphonic characters, However, this will be limited by the size of the corpus. If the corpus is too large, it will introduce matching ambiguity errors. Purely using a phrase library cannot solve the situation where a single word or word has multiple pronunciations; 3. Linguists formulate rules and then combine them Machine learning methods such as decision trees and decision trees are used to train model recognition, but it is difficult to formulate rules
Therefore, the existing polyphone pronunciation prediction accuracy rate is low

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Polyphone pronunciation prediction method and device and computer readable storage medium
  • Polyphone pronunciation prediction method and device and computer readable storage medium
  • Polyphone pronunciation prediction method and device and computer readable storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0040] The invention discloses a method for predicting the pronunciation of polyphonic characters. The scheme of converting Chinese characters containing polyphonic characters into pinyin according to the invention can be used for front-end text processing in speech recognition, and can also be used in speech synthesis and other fields that require phonetic annotation of polyphonic characters. , can be applied to electronic equipment, such as computers, servers, vehicle-mounted terminals, etc. Further, it may be applied to a scenario of all connections in a direct memory access (Direct Memory Access, DMA) link or other scenarios, which is not limited.

[0041] refer to figure 1 , the method comprises the following steps: importing the input text into a trained polyphonic character prediction model to obtain the pronunciation of the polyphonic character in the text; annotating the input text with the pronunciation of the monophonic character to obtain the pronunciation of the m...

Embodiment 2

[0073] The invention discloses a device for predicting the pronunciation of polyphonic characters. image 3 , including a polyphonic character prediction module, which is used to import the input text into the trained polyphonic character prediction model to obtain the pronunciation of the polyphonic character in the text;

[0074] The monophonic pronunciation marking module is used to mark the input text with monophonic pronunciation to obtain the monophonic pronunciation;

[0075] The pronunciation combination module is used to combine the pronunciation of monophonic characters and the pronunciation of polyphonic characters according to the text order, and output the pronunciation of the entire text.

[0076] Among them, refer to Figure 4 , the polyphonic word prediction module includes:

[0077] The input layer is used to input training texts containing polyphonic characters, and output labeled data texts;

[0078] The pre-training layer is used to input the labeled dat...

Embodiment 3

[0089] The present invention discloses a computer-readable storage medium, including a set of computer-executable instructions, which are used to execute the method for predicting the pronunciation of polyphonic characters in Embodiment 1 when the instructions are executed.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a polyphone pronunciation prediction method, relates to the technical field of computer voice processing, and aims to solve the problem of low polyphone pronunciation marking accuracy in the prior art. According to the technical scheme, the method comprises the steps: obtaining a large number of texts containing polyphone and pinyin complete spelling of the texts; trainingon a designed model by using a batch iterative training method to obtain a polyphone prediction model; in the text pronunciation labeling system, obtaining a text input by a user, predicting pronunciation of the text by using the polyphone prediction model, looking up a table to obtain single-pronunciation pinyin, and splicing and outputting pinyin corresponding to the text.. According to the method, the context information of the text is learned by using a deep neural network, the polyphone pronunciation is predicted, and the effect of improving the prediction accuracy of the polyphone pronunciation is achieved.

Description

technical field [0001] The invention relates to the technical field of computer speech processing, in particular to a method for predicting the pronunciation of polyphonic characters. Background technique [0002] Speech synthesis, a technology that allows computers to synthesize corresponding voices based on text content, enables machines to speak, and is the key to improving the human-computer interaction experience. At present, deep learning technology has also entered the field of speech synthesis and has achieved good results. The invention is used to convert Chinese text with polyphonic characters into correct pinyin, which is a key step of speech synthesis. [0003] The previous pronunciation prediction methods for polyphonic characters mainly include: 1. Directly follow the pronunciation with the highest frequency, obviously this method has a low accuracy rate; 2. Summarize the polyphonic word lexicon and corpus, and then use the phrase matching method to process po...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G10L13/02G10L13/08G06N3/04G06N3/08
CPCG10L13/02G10L13/08G06N3/08G06N3/045
Inventor 司马华鹏王培雨
Owner NANJING SILICON INTELLIGENCE TECH CO LTD