摘要
为提高音乐的发音识别和分辨能力,从而提高音乐的欣赏和评价水平,提出一种基于声纹特征提取的音乐语音信号分析和识别方法,对提取的音乐语音数据进行信号重组和特征分解,采用经验模态分解方法对音乐语音信号进行时频转换,采用自适应滤波方法进行降噪分离,实现对音乐语音信号的信息提纯和特征分解,提取提纯后的音乐语音信号的声纹特征,根据声纹特征提取结果进行音乐识别。仿真结果表明,采用该方法进行语音特征提取的准确性较好,降噪能力较强,对音乐识别的分辨能力较高。
In order to improve the pronunciation recognition music and the resolution, so as to improve the level of music appreciation and evaluation, put forward a kind of music and speech signal analysis and recognition method based on feature extraction of voice, music voice data extracted from signal reorganization and characteristic decomposi- tion, using empirical mode decomposition method for music voice signal conversion the adaptive filtering method for noise reduction, separation, decomposition of music information of speech signal purification and characteristics of voiceprint music after purification of speech signal extraction, music recognition based on the results of voiceprint ex- traction. The simulation results show that using the method of speech feature extraction accuracy better denoising abili- ty, for music recognition high resolution.
作者
余莉娟
YU Lijuan(Shangluo college shanxi province ,Shanxi Shangluo, 72600)
出处
《自动化与仪器仪表》
2018年第6期19-21,共3页
Automation & Instrumentation
关键词
声纹特征
音乐
信号
识别
voiceprint
music
signal recognition