A voice recognition method and device

Abstract

The invention discloses a voice recognition method and a voice recognition device. The method comprises steps: when voice input trigger operation is detected, a recording thread in the terminal is controlled is constantly in a starting state during a preset process during which a preset voice input method is enabled, and thus, first inputted voice information is recorded through the recording thread; and a first recognition thread is started for recognizing the first voice information to first text information through the first recognition thread. Through the technical scheme of the invention, through controlling the recording thread to be constantly in the starting state during the preset process during which the preset voice input method is enabled, the first voice information inputted during the recording process of the time can be recorded, repeated starting of the recording thread is not needed when the user repeatedly starts a recording key, the terminal is constantly in a recording state during the preset process, and a recording delay problem in a related technology can be avoided as much as possible.

Description

technical field [0001] The present invention relates to the technical field of terminals, in particular to a voice recognition method and device. Background technique [0002] At present, when using the voice input method for voice recognition, the recording is started only when the recording button is pressed, and then the recorded voice is recognized as text and then output. However, the recording scheme in the related technology corresponds to two threads, A recording thread and a recognition thread. The specific recording process is as follows: When the recording button is pressed, the recording thread and the recognition thread corresponding to the voice input method are started, and the recording thread is responsible for starting the terminal recording, and the recognition thread is responsible for fetching from the recording thread. Record the data and send it to the server for recognition. After the recognition is completed, the recognized text is obtained from the ...

AI Technical Summary

Problems solved by technology

[0002] At present, when using the voice input method for voice recognition, the recording is started only when the recording button is pressed, and then the recorded voice is recognized as text and then output. However, the recording scheme in the related technology corresponds to two threads, A recording thread and a recognition thread. The specific recording process is as follows: When the recording button is pressed, the recording thread and the recognition thread corresponding to the voice input method are started, and the recording thread is responsible for starting the terminal recording, and the recognition thread is responsible for fetching from the recording thread. Record the data and send it to the server for recognition. After the recognition is completed, the recognized text is obtained from the server and output. However, it takes a certain amount of time for the recording thread to start the terminal recording. Therefore, the recording scheme in the related technology may cause recording delay. At the same time, due to the problem of processing delay in the recognition thread, the current recognition process cannot be ended immediately when the recording button is lifted, and there is only one recognition thread. Therefore, when the user continuously lifts and presses the recording button in a short period of time, The recognition thread is likely to be unable to process the next recording process immediately because it has not yet recognized the recording data of the previous recording process, resulting in a more serious recognition lag problem or the recognition thread can only discard part of the voice information in order to reduce the recognition lag problem as much as possible. Recognition processing, which affects the user's voice input experience and brings great inconvenience to the user

Images