Text-to-speech conversion method and device, electronic device and storage medium

Abstract

The invention discloses a text-to-speech conversion method and device, an electronic device and a storage medium. The method comprises steps that a preset text normalization template matching a to-be-processed text is obtained; text normalization processing of the to-be-processed text is performed according to the matched preset text normalization template to obtain a normalization text; accordingto a word pre-segmentation template corresponding to the preset text normalization template, the word pre-segmentation information is added to the normalization text; the normalization text is segmented according to the word pre-segmentation information and a word segmentation model to obtain a word segmentation text; the word segmentation text is converted to a voice message. The method is advantaged in that word segmentation accuracy is improved, and accuracy of the synthesized speech is guaranteed.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of natural language, and in particular, relate to a text-to-speech method, device, electronic device, and storage medium. Background technique [0002] Text to Speech (TTS) is a technology that converts text into human natural language. It is widely used in car navigation broadcast, online customer service of merchants, and intelligent robot language interaction. [0003] The TTS front-end includes text normalization (Text Normalize, TN) for the input text, word segmentation, word transcription, part-of-speech prediction and pause prediction. The accuracy of word segmentation will directly affect the accuracy of subsequent phonetic transcription, part of speech, and pause prediction, thus reflecting the naturalness of the final synthesis. The word segmentation scheme in the prior art requires a dictionary and labeled training corpus. During training, the frequency of the current word ...

AI Technical Summary

Problems solved by technology

[0004] However, during the actual implementation process, the inventor found that word segmentation in the prior art is prone to errors. On the one hand, it is impossible to separate words that are not registered in the dictionary. Such words appear in many occasions such as personal names and place names, and the wrong results are correct. Great influence; on the other hand, consecutive numbers may be separated, because the combination of numbers is endless, and it is impossible to write them all into the dictionary

The error of the word segmentation system will affect the synthesis effect of the entire TTS system

Images