Acoustic model training method and device, voice recognition method and device and electronic equipment

Abstract

The invention discloses an acoustic model training method and device, a voice recognition method and device and electronic equipment. The acoustic model training method comprises the steps that: a bilingual mixed dictionary is acquired, the bilingual mixed dictionary comprises a corresponding relation between vocabularies and pronunciation annotations, and the pronunciation annotations corresponding to foreign language vocabularies in the bilingual mixed dictionary comprise at least one initial consonant and / or final; according to the bilingual mixed dictionary, pronunciation annotations corresponding to vocabularies contained in texts corresponding to training data in a training data set are determined, and the training data at least comprise bilingual mixed audio data; a pronunciation annotation sequence corresponding to the texts is obtained according to the determined pronunciation annotations, and the pronunciation annotation sequence is determined as a training label of the training data; and the acoustic model is trained according to the training data and the training label of the training data. According to the technical scheme provided by the embodiment of the invention, the acoustic model obtained by training better conforms to the pronunciation habit of Chinese when speaking foreign languages, and the accuracy of recognizing bilingual mixed voices is improved.

Description

technical field [0001] The invention relates to the technical field of speech recognition, in particular to an acoustic model training method, a speech recognition method, a device and electronic equipment. Background technique [0002] With the popularization of foreign language education and the globalization of life, people’s daily communication is often mixed with foreign language vocabulary. For example, when you want to listen to English songs, you will say "play me yesterday once more", and you will also say "you give me shut up" expression. In human-computer interaction, bilingual expressions undoubtedly bring challenges to speech recognition. [0003] At present, the method of phoneme fusion is used to solve the problem of bilingual mixed speech recognition, that is, the phonemes of Chinese and foreign languages ​​are fused together to build an acoustic model. However, this leads to a larger phone set, which multiplies the number of three-factor states, and it is ...

AI Technical Summary

Problems solved by technology

[0004] Embodiments of the present invention provide an acoustic model training method, speech recognition method, device, electronic equipment, and storage medium to solve the problem of low accuracy in recognizing bilingual mixed speech in the prior art

Images