论文部分内容阅读
提出了全局谱参数下的耳语说话人状态因子分析方法。首先,根据耳语听辨实验结果,提出导入唤醒度-愉悦度因子对说话人状态进行三级度量;其次,提取耳语音正弦模型、人耳听觉模型下的谱参数,结合其他短时频谱参量,进行轨迹跟踪并计算各参数的全局统计变量,作为特征参数来实现耳语说话人状态的分类。实验结果显示,正弦模型及人耳听觉模型的全局谱参数可将耳语说话人状态因子分类系统的准确率提高至90%。该分类方法及状态因子描述方案提供了耳语音说话人状态分析的有效途径。
The method of state factor analysis of Whisper Speaker under global spectral parameters is proposed. First of all, according to experimental results of whisper-earning and hearing, this paper proposes a three-level measurement of speaker state by introducing arousal-pleasure factor. Secondly, extracting the sinusoid model of whispered speech and spectral parameters of human auditory model, combining with other short- Track the trajectory and calculate the global statistical variables of each parameter as feature parameters to achieve the classification of whisper speaker state. The experimental results show that the global spectral parameters of the sinusoidal model and the human ear auditory model can improve the accuracy of the speaker factor state classification system to 90%. The classification method and state factor description scheme provide an effective way to analyze the state of whispered speech speaker.