Voice recognition method and device, equipment and storage medium

Abstract

The invention provides a voice recognition method, device and equipment and a storage medium, and relates to the technical field of voice recognition. According to the invention, rhythm detection is carried out on a to-be-recognized voice, tone modification is carried out on the phoneme posterior probability according to the rhythm detection result, and decoding path search is carried out according to the tone-modified phoneme posterior probability, so that the voice recognition accuracy is improved. Rhythm prediction is carried out on the to-be-recognized voice to obtain a rhythm structure ofthe to-be-recognized voice; pronunciation prediction is carried out on the to-be-recognized voice according to the acoustic characteristics of the to-be-recognized voice to obtain a plurality of phoneme posterior probabilities of the to-be-recognized voice; according to the rhythm structure, tone changing is carried out on one or more phoneme posterior probabilities in the plurality of phoneme posterior probabilities; and path search is carried out in a finite state converter according to the plurality of phoneme posteriori probabilities after tone change, and decoding is carried out to obtain a corresponding text of the to-be-recognized voice.

Description

technical field [0001] The present application relates to the technical field of speech recognition, in particular to a speech recognition method, device, equipment and storage medium. Background technique [0002] Automatic Speech Recognition (ASR) is a technology that studies how to convert human speech recognition into text. It is widely used in voice dialing, voice navigation, indoor device control, voice document retrieval, simple dictation data entry, etc. in service. [0003] In actual human pronunciation, pitch changes will occur when syllables are uttered continuously, that is, the tones of some syllables will be affected by the tones of the following tones and change. The current speech recognition technology cannot recognize the tone shift in the speech without increasing the decoding path, which leads to the inability to correctly recognize the speech with the pitch shift pronunciation. Contents of the invention [0004] Embodiments of the present application...

AI Technical Summary

Problems solved by technology

The current speech recognition technology cannot recognize the tone shift in the speech without increasing the decoding path, so the speech with the pitch shift pronunciation cannot be correctly recognized

Images