A sentiment asymmetric speaker recognition system based on a generative adversarial network

By generating high-quality sentiment spectrograms using adversarial generative networks, the problem of sentiment asymmetry in speaker recognition systems is solved, improving recognition accuracy and robustness, especially under non-neutral sentiment conditions.

CN116543774BActive Publication Date: 2026-07-24EAST CHINA UNIV OF SCI & TECH
2 Cites 0 Cited by

Patent Information

Application Number
CN202310506098.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-05-06
Publication Date
2026-07-24
Estimated Expiration
2043-05-06

AI Technical Summary

Technical Problem

Existing speaker recognition technologies suffer from inconsistent emotional distributions between training and testing speech under conditions of emotional asymmetry, leading to decreased recognition performance. Furthermore, existing methods require emotional speech data from the target speaker, which limits the effectiveness of the recognition system.

Method used

We employ adversarial generative networks to generate other sentiment spectrograms corresponding to the neutral spectrograms of registered users. By introducing cycle consistency loss, adversarial loss, speaker identity consistency loss, and sentiment classification loss, we train the generator and discriminator to generate high-quality sentiment spectrograms while preserving speaker identity information, thus expanding the spectrogram sample size of the training data.

Benefits of technology

It effectively solves the problem of inconsistent speech distribution under different emotional states, improves the robustness and accuracy of speaker recognition, and significantly improves the recognition rate, especially under non-neutral emotional conditions.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The application provides a sentiment asymmetry speaker recognition system based on a generative adversarial network, and a use process of the system comprises the following steps: firstly, converting the voice corresponding to a plurality of sentiment texts into pre-learned multi-sentiment spectrograms through pre-emphasis, framing and windowing and fast Fourier transform; secondly, training the generative adversarial network using the pre-learned multi-sentiment spectrograms, and supervising the conversion between the spectrograms of neutrality and other sentiments in the training process of the generative adversarial network through overall loss; then, converting the spectrogram of neutrality of a registered user into other sentiment spectrograms using the generative adversarial network, and obtaining the user registration multi-sentiment spectrogram by combining the spectrogram of neutrality and other sentiment of the registered user; finally, training a speaker recognition network using the user registration multi-sentiment spectrogram, calculating the speaker classification probability of the voice to be detected, and obtaining the final voiceprint recognition result. The system solves the problem of the decline of the performance of speaker recognition caused by the inconsistency between the registration and the voice sentiment in the actual application scene.
Need to check novelty before this filing date? Find Prior Art