Speech emotion recognition system and recognition method

Abstract

The invention discloses a speech emotion recognition system, including a speech preprocessing module, an emotional feature extraction module, and an affective analysis module. The input end of the speech preprocessing module is connected with the speech data, the output end of the speech preprocessing module is connected with the input end of the emotion feature extraction module, the output end of the emotion feature extraction module is connected with the input end of the emotion analysis module, and the output end of the emotion analysis module outputs an analysis recognition result. The speech pre-processing module obtains the speech signal through processing the speech data, and transmits the speech signal to the emotion feature extraction module to extract the acoustic parameters closely related to the emotion in the speech signal, and finally sends the acoustic parameters to the emotion analysis module to complete the emotion judgment. The invention also provides a speech emotion recognition method, which adds a detection means of a telephone fraud system, and can carry out multi-dimensional analysis on the speech data, and the detection accuracy rate of the system is improved by 5%.

Description

technical field [0001] The invention relates to a voice emotion recognition system and a recognition method, belonging to the technical field of voice analysis. Background technique [0002] At present, the existing telephone fraud prevention and interception systems are mainly the early warning and prevention technology based on signaling data, the fraud call early warning technology based on harmful recording comparison, and the natural person fraud call early warning technology based on intelligent voice technology. These technical routes exist. For the following problems, simple signaling analysis and recording comparison analysis do not provide much feature information, and it is difficult to give consideration to all parties in terms of accuracy and comprehensiveness. In addition, for voice topic analysis, it is necessary to accumulate voices for a period of time, which requires relatively high system voice access capabilities and processing and analysis capabilities, ...

AI Technical Summary

Problems solved by technology

It overcomes the technical defects in the traditional technical solutions that only literal information can be obtained through basic intent understanding, and it is impossible to deeply dig out abnormal information caused by changes in emotions and emotions

Images