资源列表
homework1
- 清华大学信号处理原理课程第一次实验，内容为产生固定波形以及声波的叠加-Signal processing theory, Tsinghua course first experiment, the content and the acoustic waveform to generate a fixed superposition
hmm
- Runs an HMM as a generative model generating N output vectors
mfcc
- omputes Mel Frequency Cepstrum Coefficients (MFCC)
MMSE_SA
- this code is about MMSE algorithm for speech enhancement that use SA method.
subbandspectralentropy
- 基于子带频谱墒的语音端点检测程序,程序思路清晰易懂-Based on the sub-band spectrum entropy Speech Endpoint Detection procedures, procedures clear and understandable
direct2
- 实现 直接2型滤波器的设计 以及 仿真 功能-direct two-filter design and simulation function
gllfx
- 制作一个界面，添上必需的命令按钮，分析wav文件的过零率-Analysis of zero-crossing rate wav files
t1
- decribes designa dn implementation of fractuonal dlay filter for delaying signals at input of microphone array
LMS
- LMS算法实现自适应滤波 clear close all clc N=10000 设置仿真长度 信号产生参数设定 a1=-0.195 a1=-1.5955 a2=0.95 R0=[1,a1,a2 a1,1+a2,0 a2,a1,1] p=[1,0,0] r=inv(R0)*p 计算理论自相关函数 R=[r(1),r(2) r(2),r(1)] 生成理论自相关矩阵 p1=[r(2),r(3)] 生成互相关 h=inv(R)
LPC-speech
- 对加窗后的语音帧计算lpc系数.用倒谱的方法求频谱 加海明窗的原始语音帧LPC倒谱.-Calculate lpc coefficients of the windowed speech frames. Cepstrum and spectrum the Jia Haiming window of the original speech frame LPC cepstrum.
mfcc_extraction
- 音频特征MFCC系数提取函数，包含静态12维，一阶差分和二阶差分24维，共36维，能够极大地提高音频识别的效果。-MFCC coefficient audio feature extraction functions, including static 12-dimensional, first-order and second-order differential difference dimension 24, a total of 36 dimensions, can greatly im
pitch
- 自己编的一个利用自相关法求解基音周期的程序-own addendum to the use of an auto-correlation method Pitch procedures