calpy.entropy package

Submodules

calpy.entropy.entropy module

calpy.entropy.entropy.entropy_profile(symbols, window_size=100, window_overlap=0)

Calculate the entropy profile of a list of symbols.

Args:

symbols (numpy.array or list (int)): A list of symbols. window_size (int, optional): Number of symbols per entropy window. Defaults to 100. window_overlap (int, optional): How much the entropy windows should overlap. Defaults to 0.

Returns:

numpy.array(float): The entropy profile.

calpy.entropy.entropy.entropy_profile_2D(symbols, window_size=100, window_overlap=0)

Calculate the 2D entropy profile of symbols (typically mfcc with axis 1 as time).

Args:

symbols (2D numpy.array (int)): Symbols of 2 dimensions. window_size (int, optional): Number of symbols per entropy window. Defaults to 100. window_overlap (int, optional): How much the entropy windows should overlap. Defaults to 0.

Returns:

2D numpy.array(float): The 2D entropy profile.

calpy.entropy.entropy.estimate_Gaussian(X)

Estimate the parametres of a Gaussian distribution using X

Args:

X (numpy.array (float)): Training dataset with features along axis 0, and examples along axis 1.

Returns:

Mu (numpy.array (float)): mean of the X (n by 1 dimension). Sigma2 (numpy.array (float)): variance of X (n by 1 dimension).

calpy.entropy.entropy.multivariate_Gaussion(X, Mu, Sigma2)

Computes the probability density function of multivariate Gaussian distribution.

Args:

X (1D numpy.array (float)): n by 1 feature vector. Mu (1D numpy.array (float)): n by 1 mean vector. Sigma2 (1D numpy.array (float)): n by 1 variance vector.

Returns:

p (float): probability of input X.

calpy.entropy.entropy.symbolise(pitches, eps=0.08)

Symbolose a small piece of speech segment according to pitch slopes.

Args:

pitches (numpy.array or list (float)): A list of pitches. eps (float, optional): Treshold of pitch slopes to be considered level. Defaults to tan(5 degrees).

Returns:

int: one symbol of the small piece of speech segment.

calpy.entropy.entropy.symbolise_mfcc(mfcc)

Symbolise speech according to mfcc.

Args:

mfcc (2D numpy.array (float)): A list of mfcc, axis 1 is time and axis 0 is mfcc

Returns:

symbols (numpy.array (float)): A list of symbols.

calpy.entropy.entropy.symbolise_mfcc_multidimension(mfcc)

Symbolise speech in multi-dimensional scale according to mfcc.

Args:

mfcc (2D numpy.array (float)): A list of mfcc, axis 1 is time and axis 0 is mfcc

Returns:

symbols (2D numpy.array(int)): Multi-dimensional symbols. A 2D numpy.array with the same shape as input mfcc

calpy.entropy.entropy.symbolise_pauses(pause_A, pause_B)

symbolise a conversation between two speakers into 63 patterns with their pause profiles.

Args:

pause_A (numpy.array(int)): 0-1 1D numpy integer array with 1s marking pause of speaker A. pause_B (numpy.array(int)): 0-1 1D numpy integer array with 1s marking pause of speaker B.

Returns:

symbols (numpy.array(int)): a 2D numpy.array with shape (64, pause_A.shape[0]). Axis 1 is the temporal dimension and axis 0 marks the pattern

calpy.entropy.entropy.symbolise_speech(pitches, pauses, thre=250)

Symbolise a small speech segment according to pitch and pause.

Args:

pitches (numpy.array(float)): A list of pitches. pauses (numpy.array(int)): A list of pauses. thre (float, optional): Threshold of high pitch.

Returns:

int: one symbol of the small speech segment.