Pronunciation dictionary construction method, speech recognition method and related device

Abstract

The invention provides a pronunciation dictionary construction method, a speech recognition method and a related device, and the method comprises the steps: carrying out the phoneme extraction of target audio data, and obtaining a corresponding phoneme set; wherein the target audio data is audio data covering all phonemes; according to the identification probability of each phoneme in the extracted phoneme set and a phoneme label corresponding to the target audio data, determining similar phonemes of the phonemes in the phoneme label from the phoneme set; and constructing a multi-pronunciation dictionary according to a rule of correspondingly storing the word and the standard pronunciation and the similar pronunciation corresponding to the word; wherein the standard pronunciation is composed of phonemes in the phoneme label, and the similar pronunciation is composed of similar phonemes of the phonemes in the phoneme label. The multi-pronunciation dictionary constructed through the above scheme can improve the fault tolerance and robustness of speech recognition, and further can improve the speech recognition effect.

Description

technical field [0001] The present application relates to the technical field of speech recognition, in particular to a pronunciation dictionary construction method, speech recognition method, device, equipment and storage medium. Background technique [0002] At present, mainstream commercial speech recognition systems are still based on the joint decoding framework of acoustic models and language models. The acoustic model is mainly responsible for the mapping of speech features to phonemes, and the language model combines pronunciation dictionaries to convert phoneme strings into corresponding text strings. [0003] In the pronunciation dictionary, the correspondence between the pronunciation constituted by phonemes and the text is recorded. When the pronunciation of the phoneme string recognized by the acoustic model matches a certain pronunciation in the pronunciation dictionary, the text corresponding to the phoneme string can be determined according to the corresponde...

AI Technical Summary

Problems solved by technology

[0004] Existing speech recognition schemes have poor fault tolerance. Specifically, only phoneme strings that strictly match the pronunciations in the pronunciation dictionary will be recognized as texts corresponding to the pronunciations in the pronunciation dictionary. Although this can ensure speech recognition Absolute accuracy, but it reduces the robustness of speech recognition, which is not conducive to the improvement of speech recognition effect

Images