论文部分内容阅读
根据实际应用的需求,设计了一种在浴室噪声环境中的小词汇量的语音识别系统.针对特定的浴室噪声进行分析,引入一种语音增强算法来降低噪声对识别的影响.使用基于人耳听力模型的mel频率倒谱特征参数作为语音特征参数,并结合其一阶差分参数描述语音信号特征的动态特性.针对小词汇量的实际应用需求,选用基于动态时间规整的模板匹配方法来识别.该方法可以用于其他相似的噪声环境中,通过实验验证了该方法的可行性.
According to the requirements of practical application, this paper designs a small vocal recognition system in bathroom noise environment.According to the analysis of the specific bathroom noise, a speech enhancement algorithm is introduced to reduce the noise impact on the recognition.Using human- The frequency characteristic of the mel frequency cepstrum of the hearing model is taken as the speech feature parameter and the first-order difference parameter is used to describe the dynamic characteristics of the speech signal characteristics.According to the practical application demand of small vocabulary, the template matching method based on dynamic time warping is adopted to identify. This method can be used in other similar noisy environments, and the feasibility of this method is verified by experiments.