A Noise Processing Method of the Speech Signal

摘要：

The speech signal is polluted by noises, the quality of speech signal drops, so the understandability reduces. That pre-processing the speech signal including noises, extracting the Mel-Frequency Cepstrum Coefficients that are more suitable for the human auditory characteristic is necessary for obtaining the purer speech signal and improving the recognition ratio. In experiment, we use the neural network as the recognition model,meanwhile make comparison in recognition ratio between the Linear Prediction Coefficients. The results prove the MFCC can filter the noises efficiently achieved in your work. Adhere to the line spacing given in this document.

关键词： Neural Network MeI-Frequency Cepstrum Coefficients (MFCC) Linear Prediction Coefficients (LPCC)

作者: LIN Kunhui XI Xiaojing QIAN Yi

作者单位: Software School XiaMen University XiaMen, 361005, China XiaMen Market Development and Construction Services Centre Xiamen Lakeside South street 43 floor 8,F

会议类型: 国际会议

会议名称: 第一届国际计算机新科技与教育学术会议(Proceedings of the First International Conference on Computer Science & Education ICCSE2006)

会议地点: 厦门

会议语种:英文

页码: 844-846

在线出版日期: 2006-07-27（万方平台首次上网日期，不代表论文的发表时间）