Method and system for character-to-sound conversion of multilingual mixed-language text

Abstract

The invention discloses a multilingual mixed-language text character-pronunciation conversion method and a multilingual mixed-language text character-pronunciation conversion system. The multilingual mixed-language text character-pronunciation conversion method comprises the steps of: receiving multilingual mixed-language text to be processed; determining primary language text and secondary language text in the multilingual mixed-language text; adopting unified pronunciation symbols for carrying out character-pronunciation conversion on the primary language text and the secondary language text in the multilingual mixed-language text, so as to obtain a primary language text phonetic symbol sequence and a secondary language text phonetic symbol sequence; correcting rhythm of the secondary language text phonetic symbol sequence; and connecting the primary language text phonetic symbol sequence with the corrected secondary language text phonetic symbol sequence to obtain a mixed-language text phonetic symbol sequence. By adopting the multilingual mixed-language text character-pronunciation conversion method and the multilingual mixed-language text character-pronunciation conversion system, the rhythmical sense at the joint of the primary language text phonetic symbol sequence and the secondary language text phonetic symbol sequence can be smoother.

Description

technical field [0001] The invention relates to the field of natural language processing, in particular to a method and system for converting multilingual mixed-language texts. Background technique [0002] With the increasing world economy and cultural exchanges, and the rapid development of Internet technology, the degree of global informatization has increased rapidly, resulting in a large number of new words and terms. However, not all languages ​​can fully express all new words and terms. Therefore, texts in other languages ​​often appear in one language text, that is, mixed language texts, such as "APEC was established in 1989", "Microsoft Office is a set of office software developed by Microsoft Corporation" and other Chinese mixed with English texts Condition. Generally, the main language of the mixed language text is predetermined, and the main language text occupies the main content of the mixed language text, and the non-main language text is called the secondar...

AI Technical Summary

Problems solved by technology

Obviously, the existing method is likely to cause a poor sense of rhythm at the connection between the main language and the secondary language in the mixed-language text, and the rhythm is not stable, thereby reducing the application effect

For example, in the splicing speech synthesis system of mixed language text, the splicing of the synthesized speech segments of the main language and the secondary language tends to have poor rhythm stability and abrupt rhythm changes, which seriously affects the naturalness of the mixed language text synthesized speech

Images