CAJ | 학술논문

为了提高说话人识别抗噪系统的性能，提出了将RLS自适应滤波器作为语音信号去噪的预处理器，进一步提高语音信号的信噪比，再通过Gammatone滤波器组，对去噪后的说话人语音信号进行处理，提取说话人语音信号的特征参数GFCC，进而将特征参数GFCC用于说话人识别系统中。仿真实验在高斯混合模型识别系统中进行。实验结果表明，采用这种方法应用于说话人识别抗噪系统，系统的识别率及鲁棒性都有明显的提高。
위료제고설화인식별항조계통적성능，제출료장RLS자괄응려파기작위어음신호거조적예처리기，진일보제고어음신호적신조비，재통과Gammatone려파기조，대거조후적설화인어음신호진행처리，제취설화인어음신호적특정삼수GFCC，진이장특정삼수GFCC용우설화인식별계통중。방진실험재고사혼합모형식별계통중진행。실험결과표명，채용저충방법응용우설화인식별항조계통，계통적식별솔급로봉성도유명현적제고。
In order to improve the performance of speaker recognition system in noisy environment, a RLS adaptive filter as preprocessor to remove noise is presented, to further improve the SNR of speech signal, and then through the Gammatone filter bank to deal with the speech signal after denoising, extracting the feature parameter GFCC, which is used for speaker recognition. Simulation experiment is conducted with Gaussian mixture model based speaker recognition system. The experimental results show that the proposed algorithm significantly improves system recognition rate and robustness in noisy environment.