melfcc(tuneR)
melfcc()所属R语言包:tuneR
MFCC Calculation
MFCC计算
译者:生物统计家园网 机器人LoveR
描述----------Description----------
Calculate Mel-frequency cepstral coefficients.
计算Mel频率倒谱系数。
用法----------Usage----------
melfcc(samples, sr = samples@samp.rate, wintime = 0.025,
hoptime = 0.01, numcep = 12, lifterexp = 0.6, htklifter=FALSE,
sumpower = TRUE, preemph = 0.97, dither = FALSE,
minfreq = 0, maxfreq = sr/2, nbands = 40, bwidth = 1,
dcttype = c("t2", "t1", "t3", "t4"),
fbtype = c("mel", "htkmel", "fcmel", "bark"), usecmp = FALSE,
modelorder = NULL, spec_out=FALSE, frames_in_rows=TRUE)
参数----------Arguments----------
参数:samples
Object of class "Wave".
对象类“波”。
参数:sr
Sampling rate of the signal.
的信号的采样率。
参数:wintime
Window length in sec.
窗口长度,单位为秒。
参数:hoptime
Step between successive windows in sec.
步骤之间以秒为单位的窗口。
参数:numcep
Number of cepstra to return.
数逆谱返回。
参数:lifterexp
Exponent for liftering; 0 = none;
指数liftering; 0 =无;
参数:htklifter
Use HTK sin lifter.
使用HTK罪升降机。
参数:sumpower
If sumpower=TRUE the frequency scale transformation is based on the powerspectrum, if sumpower=FALSE it is based on its squareroot (absolute value of the spectrum) and squared afterwards.
如果sumpower=TRUE的频率尺度变换是根据上powerspectrum,如果sumpower=FALSE它是基于它的平方根(绝对值的频谱)和平方之后。
参数:preemph
Apply pre-emphasis filter [1 -preemph] (0 = none).
申请预加重滤波器[1 preemph](0 =无)。
参数:dither
Add offset to spectrum as if dither noise.
添加如果抖动噪声的频谱偏移。
参数:minfreq
Lowest band edge of mel filters (Hz).
最低带边缘的梅尔过滤器(赫兹)。
参数:maxfreq
Highest band edge of mel filters (Hz).
最高带边缘的梅尔过滤器(赫兹)。
参数:nbands
Number of warped spectral bands to use.
扭曲的光谱波段使用的数量。
参数:bwidth
Width of spectral bands in Bark/Mel.
在树皮/梅尔的光谱波段宽度。
参数:dcttype
Type of DCT used - 1 or 2 (or 3 for HTK or 4 for feacalc).
使用的DCT类型 - 1或2个(或3对HTK的或4 feacalc)。
参数:fbtype
Auditory frequency scale to use: "mel", "bark", "htkmel", "fcmel"
使用的听觉频率规模:“梅尔”,“树皮”,“htkmel,fcmel
参数:usecmp
Apply equal-loudness weighting and cube-root compression (PLP instead of LPC).
等响度申请权和立方体根的压迫(PLP,而不是LPC)。
参数:modelorder
If > 0, fit a linear prediction (autoregressive-) model of this order and calculation of cepstra out of "lpcas"
如果> 0,适合一个线性预测(自回归)模型此项命令,并计算逆谱“lpcas的”
参数:spec_out
Should matrices of the power- and the auditory-spectrum be returned
如果电源和听觉频谱矩阵被返回
参数:frames_in_rows
Return time frames in rows instead of columns (original Matlab code)
返回的时限行而不是列(原始MATLAB代码)
Details
详细信息----------Details----------
Calculation of the MFCCs imlcudes the following steps: <ol> Preemphasis filtering
的MFCCs imlcudes计算以下步骤:<OL>预加重滤波
Take the absolute value of the STFT (usage of Hamming window)
取绝对值的STFT(使用汉明窗)
Warp to auditory frequency scale (Mel/Bark)
经听觉频率规模(MEL /树皮)
Take the DCT of the log-auditory-spectrum
以DCTlog听觉频谱
Return the first 'ncep' components </ol>
返回第一NCEP组件</ OL>
值----------Value----------
<table summary="R valueblock"> <tr valign="top"><td>cepstra </td> <td> Cepstral coefficients of the input signal (one time frame per row/column)</td></tr> <tr valign="top"><td>aspectrum </td> <td> Auditory spectrum (spectrum after transformation to Mel/Bark scale) of the signal</td></tr> <tr valign="top"><td>pspectrum </td> <td> Power spectrum of the input signal.</td></tr> <tr valign="top"><td>lpcas </td> <td> If modelorder > 0 the linear prediction coefficients (LPC/PLP).</td></tr> </table>
<table summary="R valueblock"> <tr valign="top"> <TD> cepstra </ TD> <TD>倒谱系数的输入信号(每行/列的一个时间框架)</ TD > </ TR> <tr valign="top"> <TD> aspectrum </ TD> <TD>听觉谱(谱改造后梅尔/巴克规模)的信号。</ TD> </ TR > <tr valign="top"> <TD> pspectrum </ TD> <TD>的输入信号的功率谱。</ TD> </ TR> <tr valign="top"> <TD> lpcas </ TD> <TD>如果modelorder > 0的线性预测系数(LPC / PLP)。</ TD> </ TR> </ TABLE>
注意----------Note----------
The following non-default values nearly duplicate Malcolm Slaney's mfcc (i.e. <pre> melfcc(d, 16000, wintime=0.016, lifterexp=0, minfreq=133.33, maxfreq=6855.6, sumpower=FALSE) </pre> =~= log(10) * 2 * mfcc(d, 16000) in the Auditory toolbox for Matlab).
以下非默认值近复制马尔科姆斯莱尼的MFCC(即<PRE> melfcc(D,16000,wintime = 0.016,lifterexp = 0,minfreq = 133.33,maxfreq = 6855.6,sumpower = FALSE)</ pre> =~= log(10) * 2 * mfcc(d, 16000)在MATLAB工具箱听觉)。
The following non-default values nearly duplicate HTK's MFCC (i.e. <pre> melfcc(d, 16000, lifterexp=22, htklifter=TRUE, nbands=20, maxfreq=8000, sumpower=FALSE, fbtype="htkmel", dcttype="t3") </pre> =~= 2 * htkmelfcc(:,[13,[1:12]]) where HTK config has "PREEMCOEF = 0.97", "NUMCHANS = 20", "CEPLIFTER = 22", "NUMCEPS = 12", "WINDOWSIZE = 250000.0", "USEHAMMING = T", "TARGETKIND = MFCC_0").
以下非默认值近HTK的MFCC(即<PRE> melfcc(D,16000复制,lifterexp = 22,htklifter = TRUE,nbands = 20,maxfreq = 8000,sumpower = FALSE,fbtype =“htkmel”,dcttype“ t3“的)</预> =~=2 * htkmelfcc(:,[13,[1:12]])其中,HTK配置有PREEMCOEF = 0.97,NUMCHANS = 20,CEPLIFTER = 22,NUMCEPS = 12, WINDOWSIZE = 250000.0,USEHAMMING = T,TARGETKIND = MFCC_0“)。
For more detail on reproducing other programs' outputs, see http://www.ee.columbia.edu/~dpwe/resources/matlab/rastamat/mfccs.html
复制其他程序的输出的更多详细信息,请参阅http://www.ee.columbia.edu/~dpwe /资源/ MATLAB / rastamat / mfccs.html
(作者)----------Author(s)----------
Sebastian Krey <a href="mailto:krey@statistik.tu-dortmund.de">krey@statistik.tu-dortmund.de</a>
参考文献----------References----------
转载请注明:出自 生物统计家园网(http://www.biostatistic.net)。
注:
注1:为了方便大家学习,本文档为生物统计家园网机器人LoveR翻译而成,仅供个人R语言学习参考使用,生物统计家园保留版权。
注2:由于是机器人自动翻译,难免有不准确之处,使用时仔细对照中、英文内容进行反复理解,可以帮助R语言的学习。
注3:如遇到不准确之处,请在本贴的后面进行回帖,我们会逐渐进行修订。
|