Speech recognition method, device and system

Abstract

The invention discloses a speech recognition method. The speech recognition method comprises the steps: first speech input is received; the received first speech input is subjected to speech rough recognition to obtain a first recognition result; second speech input is received; the received second speech input is subjected to speech rough recognition to obtain a second recognition result; according to the first recognition result, searching is conducted in an advice completion word bank to obtain advice completion words; whether the second recognition result and the advice completion words are matched or not is judged; and if yes, it is determined that the advice completion words serve as a fourth recognition result, the correlation weight of the advice completion words in the advice completion word bank is increased, and the fourth recognition result is output. Through the speech recognition method, speech recognition of subsequent input can be more quickly and more accurately achieved, the content of the advice completion word bank can be updated flexibly according to language habits and using habits of users so as to dynamically update semantic recognition, and individual needsof the users are met.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of speech recognition, and in particular to a method, device and system for speech recognition. Background technique [0002] Speech recognition technology is widely used in all aspects of the field of artificial intelligence. Artificial intelligence products such as smart home, assisted driving, and robots are inseparable from speech recognition technology. At present, although users can control artificial intelligence products through voice commands, there are still many limitations in the interaction between users and artificial intelligence products, and the content is relatively simple, which cannot yet reach or approach the communication mode between people. For example, the current speech recognition technology still has the defects of long speech input response time and low recognition accuracy. This requires an urgent solution to optimize the response time of voice input and i...

AI Technical Summary

Problems solved by technology

At present, although users can control artificial intelligence products through voice commands, there are still many limitations in the interaction between users and artificial intelligence products, and the content is relatively simple, which cannot yet reach or approach the communication mode between people

For example, the current voice recognition technology still has the defects of long voice input response time and low recognition accuracy.

Images