Speech-based role separation method and device

Abstract

The application discloses a voice-based role separation method, including: extracting feature vectors frame by frame from voice signals to obtain feature vector sequences; assigning role labels to feature vectors; using feature vectors with role labels to train deep neural network DNN models ;According to the DNN model and the hidden Markov model HMM obtained by using the feature vector training, determine the role sequence corresponding to the feature vector sequence, and output the role separation result; wherein, the DNN model is used to output the corresponding feature vector according to the input The probability of each role, HMM is used to describe the jump relationship between roles. The present application also provides a voice-based role separation device. The above-mentioned method provided by this application adopts the DNN model with strong feature extraction ability to model the speaker role, which has a stronger ability to describe the speaker than the traditional GMM, and the description of the role is more refined and accurate, so it can obtain More accurate role separation results.

Description

technical field [0001] The present application relates to the field of speech recognition, in particular to a speech-based role separation method. The present application also relates to a voice-based role separation device. Background technique [0002] Speech is the most natural way for humans to communicate, and speech recognition technology is a technology that allows machines to convert speech signals into corresponding text or commands through the process of recognition and understanding. Speech recognition is an interdisciplinary subject, and the fields involved include: signal processing, pattern recognition, probability theory and information theory, vocal mechanism and auditory mechanism, artificial intelligence, etc. [0003] In practical applications, in order to analyze speech signals more accurately, not only speech recognition is required, but also the speaker of each speech must be identified. Therefore, there is a natural need to separate speech by role. I...

AI Technical Summary

Problems solved by technology

[0005] The embodiment of the present application provides a voice-based role separation method and device to solve the problem of relatively low accuracy of existing GMM and HMM-based role separation technologies

Images