pyACA.FeatureSpectralMfccs Namespace Reference

Functions
	FeatureSpectralMfccs (X, f_s, iNumCoeffs=13)
	computes the MFCCs from the magnitude spectrum (see Slaney)
	generateDctMatrix (iNumBands, iNumCepstralCoeffs)

Function Documentation

FeatureSpectralMfccs()

FeatureSpectralMfccs	(		X,
			f_s,
			iNumCoeffs = 13 )

computes the MFCCs from the magnitude spectrum (see Slaney)

Parameters

X	spectrogram (dimension FFTLength X Observations)
f_s	sample rate of audio data
iNumCoeffs	number of coefficients to compute (default: 13)

Returns

v_mfcc: mel frequency cepstral coefficients

Definition at line 14 of file FeatureSpectralMfccs.py.

def FeatureSpectralMfccs(X, f_s, iNumCoeffs=13):
 
    isSpectrum = X.ndim == 1
    if isSpectrum:
        X = np.expand_dims(X, axis=1)
 
    # allocate memory
    v_mfcc = np.zeros([iNumCoeffs, X.shape[1]])
 
    # generate filter matrix
    H = ToolMfccFb(X.shape[0], f_s)
    T = generateDctMatrix(H.shape[0], iNumCoeffs)
 
    for n in range(0, X.shape[1]):
        # compute the mel spectrum
        X_Mel = np.log10(np.dot(H, X[:, n]) + 1e-20)
 
        # calculate the mfccs
        v_mfcc[:, n] = np.dot(T, X_Mel)
 
    return np.squeeze(v_mfcc) if isSpectrum else v_mfcc
 
 
# see function mfcc.m from Slaneys Auditory Toolbox

Here is the call graph for this function:

generateDctMatrix()

generateDctMatrix	(		iNumBands,
			iNumCepstralCoeffs )

Definition at line 38 of file FeatureSpectralMfccs.py.

def generateDctMatrix(iNumBands, iNumCepstralCoeffs):
    T = np.cos(np.outer(np.arange(0, iNumCepstralCoeffs), (2 * np.arange(0, iNumBands) + 1)) * np.pi / 2 / iNumBands)
 
    T = T / np.sqrt(iNumBands / 2)
    T[0, :] = T[0, :] / np.sqrt(2)
 
    return T

Here is the caller graph for this function: