一种基于多特征的带噪语音信号端点检测与音节分割算法

An Endpoint Detection and Syllable Separation Algorithm Based on Multi-characteristic for Noisy Speech Signal

语音信号的端点检测和音节分割直接决定语音识别率。在传统方法的基础上提取语音信号的多个特征参数,并综合利用各个参数的特性进行检测和分割,提高了端点检测和音节分割的准确度。

The endpoint detection of speech signal is an important step in speech recognition. It determines directly the recognition rate. Several characteristic parameters of speech signal are extracted based on the traditional methods for endpoint detection and syllable separation, and the accuracy of endpoint detection and syllable separation is improved.