Online real-time voice detection method and device

Abstract

The embodiment of the invention discloses an online real-time voice detection method and device. The method comprises the following steps: in a real-time voice acquisition process, generating a real-time audio clip file in each interval preset time, performing real-time scoring and comprehensive scoring on the real-time audio clip file, and immediately stopping online voice acquisition when an unqualified real-time audio clip file exists. The technical problem that a user possibly needs repeated rerecording once detection is unqualified because the whole audio file can be subjected to voice quality detection and has a certain delay after recording is stopped by trigger and the audio file is saved in the conventional method for performing voice quality detection on the audio file can be solved. Moreover, the technical problems that representation of the detection result needs to wait because the voice quality detection is performed after the audio file is saved, a large storage space isoccupied and much time is consumed by performing voice quality detection on the audio file with long duration can be solved.

Description

technical field [0001] The present application relates to the technical field of audio processing, in particular to a method and device for real-time detection of online voice. Background technique [0002] As the input of artificial intelligence capability modules such as speech recognition and voiceprint recognition, speech must be collected through specific equipment (such as microphones and supporting software applications). When collecting voice, it is often necessary to know the quality of the recording to ensure that the voice stream is available and meet the minimum standards for voice recognition and voiceprint recognition. [0003] Commonly used speech quality detection methods include methods such as VAD-SNR and MOS, all of which draw conclusions about the sound quality by analyzing the characteristics of the audio text. The determination of recording quality is usually performed at the software application layer, which requires a certain amount of computing time...

AI Technical Summary

Problems solved by technology

The judgment of recording quality is usually detected at the software application layer, which requires a certain amount of computing time and storage space.

[0004] In the existing method for voice quality detection of audio files, usually after triggering to stop recording and saving the audio file, the voice quality detection of the entire audio file can be carried out, which has a delay to a certain extent, which has caused Once the detection fails, the user may need to re-record the technical problem many times

Furthermore, the voice quality detection is performed after the audio file is saved, which will take up a large storage space, and it will also take a lot of time to perform voice quality detection on a large audio file, resulting in the technical problem of waiting for the presentation of the detection results.

Images