#!/usr/bin/env python3

# -*- coding: utf-8 -*-

"""

Created on Sat Dec 8 14:19:06 2018

@author: flash

"""

import numpy as np

import scipy.io.wavfile as wav

import os

import speechpy

import librosa

from sklearn.metrics import accuracy_score, confusion_matrix

from sklearn.model_selection import train_test_split

from sklearn import svm

from sklearn.ensemble import RandomForestClassifier

from sklearn.model_selection import cross_val_score

from sklearn.ensemble import AdaBoostClassifier

from sklearn.neural_network import MLPClassifier

from sklearn.metrics import classification_report

from keras.layers import LSTM

from keras.layers import Bidirectional

from keras.layers import TimeDistributed

from keras.models import Sequential

from keras.layers import Dense, Activation,Dropout, Flatten, Embedding

path_data = '/home/flash/Documents/IUBBooks/MLSP/Prooject/RAVDESS/Data'

#f,sr = librosa.load('/home/flash/Documents/IUBBooks/MLSP/Prooject/RAVDESS/Data/04/03-01-04-01-01-01-13.wav')

mslen = 22050

c_labels = os.listdir(path_data)

c_labels = sorted(c_labels)

print(type(c_labels))

data = []

max_fs = 0

labels = []

emotions = ['neutral','calm','happy','sad','angry','fearful','disgust','surprised']

directories = os.listdir(path_data)

print(directories)

#data_set = np.zeros((1,19240))

#dataset = []

#labels = []

#count = 0

##for d in directories:

## for file in os.listdir(os.path.join(path_data,d)):

##

## #fs,signal = wav.read(os.path.join(path_data,d,file))

## f,sr = librosa.load(os.path.join(path_data,d,file),duration=3)

##

##

## #mfcc = speechpy.feature.mfcc(signal,fs,frame_length = 0.020,frame_stride = 0.01)

## mfcc = librosa.feature.mfcc(f,sr)

## inter = mfcc.flatten()

## inter = inter.reshape((1,inter.shape[0]))

## if inter.shape[1] < 2600:

## n = 2600 - inter.shape[1]

## n_arr = np.zeros((1,n))

## inter = np.hstack((inter , n_arr))

#s = []

#m = []

#c = []

#me = []

#co = []

#to =[]

'''

feature_all=np.empty((0,193))

for d in directories:

for file in os.listdir(os.path.join(path_data,d)):

#X,sr = librosa.load('/home/flash/Documents/IUBBooks/MLSP/Prooject/RAVDESS/Data/01/03-01-01-01-01-01-01.wav',sr = None)

X,sr = librosa.load(os.path.join(path_data,d,file),duration=3)

stft = np.abs(librosa.stft(X))

#print("stft shape: ", np.shape(stft))

#s.append(np.shape(stft))

mfccs = np.mean(librosa.feature.mfcc(y=X, sr=sr, n_mfcc=40).T,axis=0)

#mfccs=mfccs.reshape(1,np.shape(mfccs)[0])

#print("mfccs shape: ", np.shape(mfccs))

#m.append(np.shape(mfccs))

chroma = np.mean(librosa.feature.chroma_stft(S=stft, sr=sr).T,axis=0)

#chroma = chroma.reshape(1,np.shape(chroma)[0])

#print("chroma shape: " , np.shape(chroma))

#c.append(np.shape(chroma))

mel = np.mean(librosa.feature.melspectrogram(X, sr=sr).T,axis=0)

# mel = mel.reshape(1,np.shape(mel)[0])

#print("mel shape:" , np.shape(mel))

#me.append(np.shape(mel))

contrast = np.mean(librosa.feature.spectral_contrast(S=stft, sr=sr).T,axis=0)

#contrast = contrast.reshape(1,np.shape(contrast)[0])

#print("contrast shape:",np.shape(contrast))

#co.append(np.shape(contrast))

tonnetz = np.mean(librosa.feature.tonnetz(y=librosa.effects.harmonic(X),sr=sr).T,axis=0)

#tonnetz = tonnetz.reshape(1,np.shape(tonnetz)[0])

#print("tonnetz shape:",np.shape(tonnetz))

#to.append(np.shape(tonnetz))

features = np.hstack([mfccs,chroma,mel,contrast,tonnetz])

#features = features.reshape(np.shape(features)[0],1)

feature_all = np.vstack([feature_all,features])

labels.append(d)

#print(feature.shape)

'''

#

import pickle

#feature_file = open('/home/flash/Documents/IUBBooks/MLSP/Prooject/RAVDESS/dec8/feature.pkl','wb')

#pickle.dump(feature_all,feature_file)

#label_file = open('/home/flash/Documents/IUBBooks/MLSP/Prooject/RAVDESS/dec8/label.pkl','wb')

#pickle.dump(labels,label_file)

f2 = open('feature.pkl','rb')

feature_all = pickle.load(f2)

f3 = open('label.pkl','rb')

labels = pickle.load(f3)

from copy import deepcopy

y = deepcopy(labels)

for i in range(len(y)):

y[i] = int(y[i])

#from sklearn.preprocessing import OneHotEncoder

#enc = OneHotEncoder(handle_unknown='ignore')

#labels =np.asarray(labels)

#labels = labels.reshape(-1,1)

#enc.fit(labels)

n_labels = len(y)

n_unique_labels = len(np.unique(y))

one_hot_encode = np.zeros((n_labels,n_unique_labels))

f = np.arange(n_labels)

for i in range(len(f)):

one_hot_encode[f[i],y[i]-1]=1

X_train,X_test,y_train,y_test = train_test_split(feature_all,one_hot_encode,test_size = 0.3,random_state=20)

model = Sequential()

model.add(Dense(X_train.shape[1],input_dim =X_train.shape[1],init='normal',activation ='relu'))

model.add(Dense(400,init='normal',activation ='relu'))

model.add(Dropout(0.2))

model.add(Dense(200,init='normal',activation ='relu'))

model.add(Dropout(0.2))

model.add(Dense(100,init='normal',activation ='relu'))

model.add(Dropout(0.2))

model.add(Dense(y_train.shape[1],init='normal',activation ='softmax'))

model.compile(loss = 'categorical_crossentropy',optimizer='adadelta',metrics=['accuracy'])

model.fit(X_train,y_train,nb_epoch=200,batch_size = 5,verbose=1)

model.evaluate(X_test,y_test)

mlp_model = model.to_json()

with open('mlp_model_relu_adadelta.json','w') as j:

j.write(mlp_model)

model.save_weights("mlp_relu_adadelta_model.h5")

y_pred_model1 = model.predict(X_test)

y2 = np.argmax(y_pred_model1,axis=1)

y_test2 = np.argmax(y_test , axis = 1)

count = 0

for i in range(y2.shape[0]):

if y2[i] == y_test2[i]:

count+=1

# json_file = open('model.json', 'r')

# loaded_model_json = json_file.read()

# json_file.close()

# loaded_model = model_from_json(loaded_model_json)

# # load weights into new model

# loaded_model.load_weights("model.h5")

model2 = Sequential()

model2.add(Dense(X_train.shape[1],input_dim =X_train.shape[1],init='normal',activation ='relu'))

model2.add(Dense(400,init='normal',activation ='tanh'))

model2.add(Dropout(0.2))

model2.add(Dense(200,init='normal',activation ='tanh'))

model2.add(Dropout(0.2))

model2.add(Dense(100,init='normal',activation ='sigmoid'))

model2.add(Dropout(0.2))

model2.add(Dense(y_train.shape[1],init='normal',activation ='softmax'))

model2.compile(loss = 'categorical_crossentropy',optimizer='adadelta',metrics=['accuracy'])

model2.fit(X_train,y_train,nb_epoch=200,batch_size = 5,verbose=1)

model2.evaluate(X_test, y_test)

# mlp_model2 = model2.to_json()

# with open('mlp_model_tanh_adadelta.json','w') as j:

# j.write(mlp_model2)

# model2.save_weights("mlp_tanh_adadelta_model.h5")

y_pred_model2 = model2.predict(X_test)

y22 = np.argmax(y_pred_model2,axis=1)

y_test22 = np.argmax(y_test , axis = 1)

count = 0

for i in range(y22.shape[0]):

if y22[i] == y_test22[i]:

count+=1

from xgboost import XGBClassifier

X_train2,X_test2,y_train2,y_test2 = train_test_split(feature_all,y,test_size = 0.3,random_state=20)

model3 = XGBClassifier()

model3.fit(X_train2,y_train2)

model3.evals_result()

cross_val_score(model3, X_train2, y_train2, cv=5)

y_pred3 = model3.predict(X_test)

count = 0

for i in range(y_pred3.shape[0]):

if y_pred3[i] == y_test2[i]:

count+=1

# clf = RandomForestClassifier(n_estimators=60,max_features=8,max_depth=None,min_samples_split=3,bootstrap=True,random_state=35)

# clf = clf.fit(X_train, y_train)

# #scores = cross_val_score(clf, X_train, y_train, cv=5)

# #print(scores.mean())

# y_pred = clf.predict(X_test)

# for i in range(np.shape(y_test))

X,sr = librosa.load('/home/flash/Documents/IUBBooks/MLSP/Prooject/MLSP_Audio/chunk-00.wav',sr = None)

stft = np.abs(librosa.stft(X))

#print("stft shape: ", np.shape(stft))

mfccs = np.mean(librosa.feature.mfcc(y=X, sr=sr, n_mfcc=40),axis=1)

chroma = np.mean(librosa.feature.chroma_stft(S=stft, sr=sr).T,axis=0)

mel = np.mean(librosa.feature.melspectrogram(X, sr=sr).T,axis=0)

contrast = np.mean(librosa.feature.spectral_contrast(S=stft, sr=sr,fmin=0.5*sr* 2**(-6)).T,axis=0)

tonnetz = np.mean(librosa.feature.tonnetz(y=librosa.effects.harmonic(X),sr=sr*2).T,axis=0)

features = np.hstack([mfccs,chroma,mel,contrast,tonnetz])

#features = features.reshape(np.shape(features)[0],1)

feature_all = np.vstack([feature_all,features])

print(features.shape)

x_chunk = np.array(features)

x_chunk = x_chunk.reshape(1,np.shape(x_chunk)[0])

y_chunk_model1 = model.predict(x_chunk)

index = np.argmax(y_chunk_model1)

print('Emotion:',emotions[index])

feature_all2=np.empty((0,193))

X2,sr = librosa.load('/home/flash/Documents/IUBBooks/MLSP/Prooject/MLSP_Audio/chunk-02.wav',sr = None)

stft2 = np.abs(librosa.stft(X2))

#print("stft shape: ", np.shape(stft))

mfccs2 = np.mean(librosa.feature.mfcc(y=X2, sr=sr, n_mfcc=40),axis=1)

chroma2 = np.mean(librosa.feature.chroma_stft(S=stft2, sr=sr).T,axis=0)

mel2 = np.mean(librosa.feature.melspectrogram(X2, sr=sr).T,axis=0)

contrast2 = np.mean(librosa.feature.spectral_contrast(S=stft2, sr=sr,fmin=0.5*sr* 2**(-6)).T,axis=0)

tonnetz2 = np.mean(librosa.feature.tonnetz(y=librosa.effects.harmonic(X2),sr=sr*2).T,axis=0)

features2 = np.hstack([mfccs2,chroma2,mel2,contrast2,tonnetz2])

#features = features.reshape(np.shape(features)[0],1)

feature_all2 = np.vstack([feature_all2,features2])

print(features2.shape)

x_chunk2 = np.array(features2)

x_chunk2 = x_chunk2.reshape(1,np.shape(x_chunk2)[0])

y_chunk2_model1 = model.predict(x_chunk2)

index2 = np.argmax(y_chunk2_model1)

print('Emotion:',emotions[index2])