Voice synthesis method and device, computer device and computer readable storage medium

Abstract

The invention discloses a text conversion method and device, a computer device and a computer readable storage medium. The method comprises the steps of obtaining a to-be-converted text; performing handwritten word recognition on the text to be converted to obtain a recognition result, wherein the recognition result is that the to-be-converted text contains handwritten words, and recognizing the handwritten words in the to-be-converted text by adopting an extreme gradient lifting model; and obtaining a target conversion text corresponding to the to-be-converted text according to a recognitionresult output by the extreme gradient lifting model. Compared with deep learning, the method is high in recognition speed and high in recognition accuracy.

Description

technical field [0001] The present invention relates to the technical field of text conversion, in particular to a text conversion method, device, computer equipment and storage medium. Background technique [0002] Text regularization refers to converting handwritten words in the text into spoken words, such as converting the text "today is October 19, 2019" into "today is October 19, 2019", In this example, "2019", "10", "19" are words in handwritten form, while "2019", "Ten" and "Nineteen" are the words of "2019", "10", "19" Oral expression. [0003] Most of the existing text regularization methods are methods based on linguistic rules and deep learning. Linguistic-based methods need to exhaustively enumerate possible situations as much as possible, which is very time-consuming and inefficient; methods based on deep learning require Input the handwritten text into the deep learning model, and you can get the spoken text output by the deep learning model. However, the re...

AI Technical Summary

Problems solved by technology

[0003] Most of the existing text regularization methods are methods based on linguistic rules and deep learning. Linguistic-based methods need to exhaustively enumerate possible situations as much as possible, which is very time-consuming and inefficient; methods based on deep learning require Input the handwritten text into the deep learning model, and you can get the spoken text output by the deep learning model. However, the recognition speed of the deep learning model is slow, and if the training corpus is insufficient, the recognition accuracy of the deep learning model will be relatively low.

Images