【作者】 芮瑞； 鲍长春；

【Author】 RUI Rui BAO Chang-chun Speech and Audio Signal Processing Lab,School of Electronic Information and Control Engineering, Beijing University of Technology,Beijing 100124,China

【机构】 北京工业大学电子信息与控制工程学院语音与音频信号处理实验室；

【摘要】 本文结合贝叶斯决策理论和统计学习思想,将概率神经网络(PNN)应用于音频分类,选用较少的音频特征将音频信号分为语音、音乐和静音三种类型,最后分析了PNN的分类性能及其泛化能力。实验表明,本文算法分类效果良好,分类准确率可以达到98.67%。更多还原

【Abstract】 Combining the Bayesian strategy and statistical learning,a method based on Probabilistic Neural Network(PNN) is applied in the audio classification in this paper.The audio signal is divided into three types including speech,music and silence with few feature of audio.Finally,the performance of classification and universality of PNN is analyzed.The result of experiment indicates that the algorithm has a good performance and classification accuracy is close to 98.67%.更多还原