一种新颖的语言/音乐分割与分类方法

语音/音乐区分是音频高效编码、音频检索、自动语音识别等音频处理和分析的重要步骤。本文提出一种新颖的语音/音乐分割与分类方法,首先根据相邻帧间的均方能量差异检测音频的变化点,实现分割;然后对音频段提取低带能量方差比、倒谱能量调制、熵调制等八维特征,用人工神经网络做分类。实验结果显示,本文算法和特征具有很高的分割准确率和分类正确率。

A Novel Speech/Music Segmentation and Discrimination Algorithm

Speech and music discrimination is an important step of efficient audio coding and content-based audio retrieval as well as automatic speech recognition and audio analysis /treatment.In this paper,we propose a novel speech/music segmentation and discrimination algorithm.First,change detection is performed based on the difference of root mean squares between neighbor frames to implement speech/music segmentation,then eight features such as the variance of low-band energy ratio,cepstral energy modulation,entr...