Method, system and device for voiceprint identification based on video and spectrogram

Abstract

The invention discloses a video and speech spectrum based voiceprint identification method. The video and speech spectrum based voiceprint identification method comprises: acquiring a speech spectrumcorresponding to an audio file, wherein the audio file corresponds to a video file; establishing associations with time on the video file and the speech spectrum so as to position the same time pointin the other end of the video file and the speech spectrum while any time point is selected from any one of the video file and the speech spectrum, and obtaining a corresponding video and formant according to the associations; and identifying a voiceprint according to the video and the formant so as to determine the identity of a person to be identified. The speech spectrum of the audio file corresponding to the video file is obtained, the associations are established in the video file and the speech spectrum, and then when any time point is selected from any one of the video file and the speech spectrum, the same time point is positioned in the other one of the video file and the speech spectrum; and the video and the formant are obtained, and mouth shapes and expression when the person to be identified speaks are observed to increase the identification basis. The invention also provides a system and device having the above advantages, and a computer readable storage media.

Description

technical field [0001] The invention relates to the field of speech recognition, in particular to a voiceprint identification method, system, device and computer-readable storage medium based on video and spectrogram. Background technique [0002] The voice of each person is different, and the voice of a person is like a fingerprint of a person, with the characteristics of "everyone is different". Especially when a person is an adult, the pronunciation organs have matured, language habits have been formed, and the pronunciation is stable except under special circumstances such as the influence of diseases. And because everyone's physiological structure, living environment and other factors are different, people's voice is specific. Therefore, it is an important science and technology to carry out personal identification through voiceprint identification. [0003] The existing identification method uses the shape and trend of the formant on the spectrogram as the most impor...

AI Technical Summary

Problems solved by technology

[0004] The object of the present invention is to provide a voiceprint identification method, system, device and computer-readable storage medium based on video and spectrogram, so as to solve the problem in the prior art by using the voice corresponding to the spectrogram to identify, and the accuracy of the identification result. not high problem

Images