Speech recognition method and system

Abstract

The invention discloses a speech recognition method and system. The speech recognition method comprises the following steps of: constructing a primary decoding recognition network; searching the received speech signal in the primary decoding recognition network to obtain a primary decoding candidate option; if the primary decoding candidate option meets primary decoding conditions, outputting an optimal recognition result obtained by the primary decoding; if the primary decoding candidate option does not meet the primary decoding conditions, carrying out optimized processing on the primary decoding candidate option according to a primary decoding delay and the decoding credibility; constructing a secondary decoding recognition network according to the optimized primary decoding candidate option; carrying out secondary decoding on the received speech signal in the secondary decoding recognition network to obtain a secondary decoding candidate option; and outputting an optimal recognition result obtained by secondary decoding. According to the speech recognition method and system disclosed by the invention, the decoding efficiency and accuracy under different conditions can be balanced and the decoding performances of the system can be optimized by dynamically optimizing the search network twice.

Description

technical field [0001] The present invention relates to the technical field of intelligent information processing, and more specifically, to a voice recognition method and system. Background technique [0002] It has become an urgent need for the application and development of information technology to realize humanized and intelligent effective interaction between man and machine, and to build an efficient and natural human-machine communication environment. As the most natural and efficient means of interaction, online speech recognition applications such as sending microblogs and writing text messages have received more and more attention on small-screen mobile devices. In online applications, users have higher requirements for system response speed, and they hope to get the recognition result without obvious delay after the voice input ends. The real-time performance of a speech recognition system has become one of the most important indicators for measuring system perf...

AI Technical Summary

Problems solved by technology

If a larger N value is set, a higher accuracy rate can be guaranteed, but because more historical paths are retained in decoding, the efficiency of one-pass decoding is reduced and the search space of two-pass decoding is increased.

Performing a relatively complex two-pass decoding on this search space often takes too long, especially if the one-pass decoding has already caused a large delay, it will bring a greater delay

On the contrary, if the N value is set smaller, it may lead to the premature pruning of the optimal path without generating correct recognition results, which will affect the accuracy of decoding.

It can be seen that this method cannot well solve the balance problem of recognition rate and recognition efficiency.

Images