PNCC_M
所属分类:matlab编程
开发工具:matlab
文件大小:515KB
下载次数:20
上传日期:2018-12-10 22:16:45
上 传 者:
wingingyy
说明: PNCC特征值提取,更新了MFCC特征参数,对语音识别机器学习的性能均有提升作用。
(PNCC feature extraction updates MFCC feature parameters and improves the performance of speech recognition machine learning.)
文件列表:
PNCC_M (0, 2018-12-10)
PNCC_M\ComputeFilterResponse.m (7667, 2010-11-08)
PNCC_M\DemoBatch.m (4079, 2012-04-07)
PNCC_M\fft2melmx.m (4866, 2010-11-05)
PNCC_M\NormalizeFilterGain.m (464, 2010-11-05)
PNCC_M\PNCC_IEEETran.m (10896, 2012-04-07)
PNCC_M\PNCC_IEEETranForDemo.m (10864, 2012-04-07)
PNCC_M\sb01_Clean.sph (165888, 2007-08-30)
PNCC_M\sb01_Music_05dB.sph (165888, 2007-06-29)
PNCC_M\sb01_Street_05dB.sph (165888, 2008-09-08)
PNCC_M\sb01_White_05dB.sph (165888, 2007-06-29)
Programmed by Chanwoo Kim for IEEE Transcation on Speech, Audio, and Language Processing
Nov. 17, 2010
1) Just Run PNCC_IEEETran('outFile', 'inPutFile')
IMPORTANT : The input is assumed in the single-Channel sphere NIST format.
The sampling rate should be "16 kHz".
It does not check the header, but just skips the header.
We used this program in getting result in IEEE Transcation paper and ICASSP 2012 paper
2) Output is a feature in Sphinx format
3) To see the spectrogram demo
Type DemoBatch.m in Matlab
It will launch three figures:
a) spectrogram,
b) spectrogram using log nonliearity (This one is still better than MFCC in our recognition experiment)
c) gammatone weighting, PNCC (This one is much better than b))
In each figure you will sub-figures obtained under different noisy conditions. If the spectrogram from the noisy
condition is closer to the top subfigure in each figure, then we consider that feature to be more robust
近期下载者:
相关文件:
收藏者: