MFCC for Audio Cats and Dogs

mfcc_for_cat_dog.py

# import library
import librosa
import numpy as np

# define the file name
wav_name = 'cat_1.wav'

# define the length of features
max_len = 20000

# load the wav file
file_data, file_rate = librosa.load(wav_name)

# get mfcc in 2D
mfcc_2D = librosa.feature.mfcc(y=file_data, sr=file_rate, n_mfcc=40)

# convert mfcc in 2D to mfcc in 1D
mfcc_1D = mfcc_2D.flatten()

# pad mfcc so that all files have features in the same length
mfcc_final = np.pad(mfcc_1D, (0, max_len - len(mfcc_1D)), 'constant')

print(mfcc_2D.shape, mfcc_1D.shape, mfcc_final.shape)