Information query method and system based on voice recognition, equipment and storage medium

Abstract

The invention discloses an information query method and system based on voice recognition, computer equipment and a computer readable storage medium. The method comprises the following steps of an encoding step, inputting audio data, performing feature value extraction by adopting a transformer encoder, and outputting a two-dimensional feature value sequence; a decoding step, based on the two-dimensional feature value sequence, adopting a transform and n-gram combined decoder to carry out streaming voice decoding and recognition, after screening and outputting M results of first text sorting, based on the two-dimensional feature value sequence and the M results of first matched text sorting, adopting a conformer model to carry out non-streaming voice decoding and recognition, outputting N results of second text sorting, N and M being positive integers larger than or equal to 1, and N being smaller than or equal to M; and the assignment weighting step is used for performing assignment weighting based on a hot word dictionary in the text sorting result output in the decoding step, and outputting an optimal query result. According to the method, through adding the language model and the hot word weight function, accuracy of speech recognition is improved.

Description

technical field [0001] The present application relates to the field of information query, in particular to a speech recognition-based information query method, system, computer equipment and computer-readable storage medium. Background technique [0002] At present, with the development of speech recognition technology, it can be found that speech recognition is highly accepted by users, easy to operate, does not involve user privacy issues, and it is more convenient to promote speech recognition related applications. [0003] From the scene, speech recognition can be divided into streaming speech recognition and non-streaming speech recognition. Non-streaming speech recognition (offline recognition) means that the model recognizes after the user has finished speaking a sentence or a paragraph, while streaming speech recognition means that the model performs speech recognition synchronously while the user is still speaking. Because of its low latency, streaming speech recog...

AI Technical Summary

Problems solved by technology

[0007] However, the unified model of streaming and non-streaming speech recognition lacks specific vocabulary and speech recognition under different contexts. Special words cannot be increased in weight to improve the recognition accuracy of special words

Images