A simple but efficient real-time Voice Activity Detection algorithm
A simple but efficient real-time Voice Activity Detection algorithm
复制标题
一种简单但高效的实时语音活动检测算法
DOI：
--
复制
发表时间：
2009
期刊：
影响因子：
--
通讯作者：
M. Homayounpour
中科院分区：
文献类型：
--
作者：
M. Moattar;M. Homayounpour
语音活动检测（VAD）是所有语音和音频处理应用中非常重要的前端处理。大多数（如果不是所有）语音/音频处理方法的性能关键取决于语音活动检测的性能。理想的语音活动检测器需要独立于应用区域和噪声条件，并且在真实的应用中具有最少的参数调整。本文提出了一种接近理想的语音检测算法，与已有的一些方法相比，该算法不仅易于实现，而且对噪声具有较强的鲁棒性。所提出的方法使用短期特征，如频谱平坦度（SF）和短期能量。这有助于该方法适用于在线处理任务。所提出的方法进行了评估的几个语音语料库加性噪声，并与一些最近提出的算法进行比较。实验表明，在各种噪声条件下，令人满意的性能。
Voice Activity Detection (VAD) is a very important front end processing in all Speech and Audio processing applications. The performance of most if not all speech/audio processing methods is crucially dependent on the performance of Voice Activity Detection. An ideal voice activity detector needs to be independent from application area and noise condition and have the least parameter tuning in real applications. In this paper a nearly ideal VAD algorithm is proposed which is both easy-to-implement and noise robust, comparing to some previous methods. The proposed method uses short-term features such as Spectral Flatness (SF) and Short-term Energy. This helps the method to be appropriate for online processing tasks. The proposed method was evaluated on several speech corpora with additive noise and is compared with some of the most recent proposed algorithms. The experiments show satisfactory performance in various noise conditions.