Audio-Data-Analysis-CNN.py

import pandas as pd
import numpy as np
from numpy import argmax
import matplotlib.pyplot as plt
%matplotlib inline
import librosa
import librosa.display
import IPython.display
import random
import warnings
import os
from PIL import Image
import pathlib
import csv
# sklearn Preprocessing
from sklearn.model_selection import train_test_split
#Keras
import keras
import warnings
warnings.filterwarnings('ignore')
from keras import layers
from keras.layers import Activation, Dense, Dropout, Conv2D, Flatten, MaxPooling2D, GlobalMaxPooling2D, GlobalAveragePooling1D, AveragePooling2D, Input, Add
from keras.models import Sequential
from keras.optimizers import SGD

genres = 'blues classical country disco hiphop jazz metal pop reggae rock'.split()
for g in genres:
    pathlib.Path(f'img_data/{g}').mkdir(parents=True, exist_ok=True)
    for filename in os.listdir(f'./drive/My Drive/genres/{g}'):
        songname = f'./drive/My Drive/genres/{g}/{filename}'
        y, sr = librosa.load(songname, mono=True, duration=5)
        print(y.shape)
        plt.specgram(y, NFFT=2048, Fs=2, Fc=0, noverlap=128, cmap=cmap, sides='default', mode='default', scale='dB');
        plt.axis('off');
        plt.savefig(f'img_data/{g}/{filename[:-3].replace(".", "")}.png')
        plt.clf()
        
        
        
import split_folders
# To only split into training and validation set, set a tuple to `ratio`, i.e, `(.8, .2)`.
split_folders.ratio('./img_data/', output="./data", seed=1337, ratio=(.8, .2)) # default values

from keras.preprocessing.image import ImageDataGenerator
train_datagen = ImageDataGenerator(
        rescale=1./255, # rescale all pixel values from 0-255, so aftre this step all our pixel values are in range (0,1)
        shear_range=0.2, #to apply some random tranfromations
        zoom_range=0.2, #to apply zoom
        horizontal_flip=True) # image will be flipper horiz
test_datagen = ImageDataGenerator(rescale=1./255)


training_set = train_datagen.flow_from_directory(
        './data/train',
        target_size=(64, 64),
        batch_size=32,
        class_mode='categorical',
        shuffle = False)
test_set = test_datagen.flow_from_directory(
        './data/val',
        target_size=(64, 64),
        batch_size=32,
        class_mode='categorical',
        shuffle = False )
        
model = Sequential()
input_shape=(64, 64, 3)
#1st hidden layer
model.add(Conv2D(32, (3, 3), strides=(2, 2), input_shape=input_shape))
model.add(AveragePooling2D((2, 2), strides=(2,2)))
model.add(Activation('relu'))
#2nd hidden layer
model.add(Conv2D(64, (3, 3), padding="same"))
model.add(AveragePooling2D((2, 2), strides=(2,2)))
model.add(Activation('relu'))
#3rd hidden layer
model.add(Conv2D(64, (3, 3), padding="same"))
model.add(AveragePooling2D((2, 2), strides=(2,2)))
model.add(Activation('relu'))
#Flatten
model.add(Flatten())
model.add(Dropout(rate=0.5))
#Add fully connected layer.
model.add(Dense(64))
model.add(Activation('relu'))
model.add(Dropout(rate=0.5))
#Output layer
model.add(Dense(10))
model.add(Activation('softmax'))
model.summary()


epochs = 200
batch_size = 8
learning_rate = 0.01
decay_rate = learning_rate / epochs
momentum = 0.9
sgd = SGD(lr=learning_rate, momentum=momentum, decay=decay_rate, nesterov=False)
model.compile(optimizer="sgd", loss="categorical_crossentropy", metrics=['accuracy'])

model.fit_generator(
        training_set,
        steps_per_epoch=100,
        epochs=50,
        validation_data=test_set,
        validation_steps=200)
        
        
#Model Evaluation
model.evaluate_generator(generator=test_set, steps=50)
#OUTPUT
[1.704445120342617, 0.33798882681564246]

test_set.reset()
pred = model.predict_generator(test_set, steps=50, verbose=1)

predicted_class_indices=np.argmax(pred,axis=1)

labels = (training_set.class_indices)
labels = dict((v,k) for k,v in labels.items())
predictions = [labels[k] for k in predicted_class_indices]
predictions = predictions[:200]
filenames=test_set.filenames

print(len(filename, len(predictions)))
# (200, 200)

results=pd.DataFrame({"Filename":filenames,
                      "Predictions":predictions},orient='index')
results.to_csv("prediction_results.csv",index=False)