Audio data processing method and device for speech recognition

Abstract

The embodiment of the invention discloses an audio data processing method, device and equipment for voice recognition and a storage medium; the method comprises the steps: determining whether an audio bit stream stored in a buffering region is greater than the maximum detection length or not when the audio bit stream is greater than the target detection length; and determining a corresponding endpoint detector to perform endpoint detection on the audio bit stream according to a result of whether the determined audio bit stream is greater than the maximum detection length, wherein the endpoint detectors comprise a silence descriptor detector, a first endpoint detector based on a long-term network model and a second endpoint detector based on a short-term network model; and processing the audio bit stream according to the endpoint detection result to obtain an audio sample for speech recognition. According to the scheme, the problem that the recognition performance is reduced due to the fact that the continuous speech segments are segmented into different parts for speech recognition is avoided, the speech region cannot be damaged, and the speech recognition efficiency and accuracy are remarkably improved.

Description

technical field [0001] The embodiments of the present application relate to the computer field, and in particular to an audio data processing method and device for speech recognition. Background technique [0002] When an audio stream is used for audio transmission, the audio stream needs to be divided for speech recognition. Between the client and the server, audio stream transmission methods mainly include audio frame transmission and audio data block transmission. [0003] In the transmission method using audio frames, the audio frame sequence is sent from the audio sending end to the audio receiving end, and the receiving end decodes the audio frame sequence, and inputs it to the speech recognizer one by one for speech recognition. This method cannot realize batch processing, resulting in Speech recognition is less efficient. [0004] In the transmission method using audio data blocks, each audio data block is transmitted between the client and the server as an audio s...

AI Technical Summary

Problems solved by technology

In the prior art, the audio stream is divided into fixed-length audio segments, such as figure 1 as shown, Figure 1a It is a schematic diagram of audio data segmentation in the prior art, which divides the audio stream at intervals of 20s. This division method is less flexible and may cause continuous voice segments to be cut during voice recognition. points, thus reducing the efficiency of speech recognition

Images