CNN_Music_Generation.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"CNN_Music_Generation.ipynb","provenance":[],"collapsed_sections":[],"mount_file_id":"1isRib4CvJWPIH5_7V7-CR5Dnpkocba_t","authorship_tag":"ABX9TyNbHRxv73SP7ZlDL614mU1u"},"kernelspec":{"name":"python3","display_name":"Python 3"},"accelerator":"GPU"},"cells":[{"cell_type":"markdown","metadata":{"id":"Bew3rK3qyUDa","colab_type":"text"},"source":["Setting the seed for reproducable results"]},{"cell_type":"code","metadata":{"id":"gGmf4yyDRQwr","colab_type":"code","colab":{}},"source":["%tensorflow_version 1.x"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"mfJK-fwTwxfB","colab_type":"code","colab":{}},"source":["from numpy.random import seed\n","seed(1)\n","from tensorflow import set_random_seed\n","set_random_seed(2)\n","from collections import Counter\n","import matplotlib.pyplot as plt"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"Ad93ifFw64j3","colab_type":"text"},"source":["Importing Library"]},{"cell_type":"code","metadata":{"id":"HIGoSUTz66lK","colab_type":"code","colab":{}},"source":["#dealing with midi files\n","from music21 import * \n","#array processing\n","import numpy as np     \n","import os\n","#random number generator\n","import random         \n","#keras for building deep learning model\n","from keras.layers import * \n","from keras.models import *\n","import keras.backend as K\n","from keras.callbacks import *\n","\n","from sklearn.model_selection import train_test_split"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"GhJnkk2y7U92","colab_type":"text"},"source":["Function for reading midi file"]},{"cell_type":"code","metadata":{"id":"kqyvwxe_7Xgx","colab_type":"code","colab":{}},"source":["def read_midi(file):\n","  notes=[]\n","  notes_to_parse = None\n","  #parsing a midi file\n","  midi = converter.parse(file)\n","  #grouping based on different instruments\n","  s2 = instrument.partitionByInstrument(midi)\n","  #Looping over all the instruments\n","  if s2 == None:\n","    print(\"NoneType !\")\n","    return None\n","  print(\"\")\n","  for part in s2.parts:\n","    #select elements of only piano\n","    if 'Piano' in str(part):\n","      notes_to_parse = part.recurse()\n","    else:\n","      notes_to_parse = None\n","    \n","    #finding whether a particular element is note or a chord\n","    if notes_to_parse != None:\n","      for element in notes_to_parse:\n","        if isinstance(element, note.Note):\n","          notes.append(str(element.pitch))\n","        elif isinstance(element, chord.Chord):\n","          notes.append('.'.join(str(n) for n in element.normalOrder))\n","\n","  return notes"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"JKyKj17k8fKK","colab_type":"code","outputId":"80b7d28f-eca0-4331-9705-627af51cbc9d","executionInfo":{"status":"ok","timestamp":1587370283092,"user_tz":240,"elapsed":3569,"user":{"displayName":"Sadman Sakif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhFzTKph8jrvXjeSmjPv-AWjlCXPOb8wSzBHYRB7w=s64","userId":"15126335579010581692"}},"colab":{"base_uri":"https://localhost:8080/","height":32}},"source":["from google.colab import drive \n","drive.mount('/content/drive', force_remount= True)"],"execution_count":54,"outputs":[{"output_type":"stream","text":["Mounted at /content/drive\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"9BXaGLCo8uWL","colab_type":"code","colab":{}},"source":["drive_path    = 'drive/My Drive/ECE496 CNN/'\n","data_in_path  = 'Trial/'\n","output_path ='Output_midi/'\n","\n","midi_dir = drive_path + data_in_path\n","out_dir = drive_path + output_path"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"kls1vCdF9zXA","colab_type":"text"},"source":["Reading midi files from directory"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"EFucZV69uNtU","outputId":"287a7669-d6c8-41fe-d46f-f8e63249adf7","executionInfo":{"status":"ok","timestamp":1587370391552,"user_tz":240,"elapsed":103827,"user":{"displayName":"Sadman Sakif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhFzTKph8jrvXjeSmjPv-AWjlCXPOb8wSzBHYRB7w=s64","userId":"15126335579010581692"}},"colab":{"base_uri":"https://localhost:8080/","height":1000}},"source":["#read all the filenames\n","#files=[i for i in \"drive/My Drive/ECE496 CNN/Dataset\"]\n","#reading each midi file\n","all_notes=[]\n","for file_name in sorted(os.listdir(\"drive/My Drive/ECE496 CNN/Trial\")):\n","  print(file_name)\n","  all_notes.append(np.array(read_midi(midi_dir + file_name)))\n","\n","#notes and chords of all the midi files\n","notes = [element for notes in all_notes for element in notes]"],"execution_count":56,"outputs":[{"output_type":"stream","text":["001.midi\n","\n","002.midi\n","\n","003.midi\n","\n","004.midi\n","\n","005.midi\n","\n","006.midi\n","\n","007.midi\n","\n","008.midi\n","\n","009.midi\n","\n","010.midi\n","\n","011.midi\n","\n","012.midi\n","\n","013.midi\n","\n","014.midi\n","\n","015.midi\n","\n","016.midi\n","\n","017.midi\n","\n","018.midi\n","\n","019.midi\n","\n","020.midi\n","\n","021.midi\n","\n","022.midi\n","\n","023.midi\n","\n","024.midi\n","\n","025.midi\n","\n","026.midi\n","\n","027.midi\n","\n","028.midi\n","\n","029.midi\n","\n","030.midi\n","\n","031.midi\n","\n","032.midi\n","\n","033.midi\n","\n","034.midi\n","\n","035.midi\n","\n","036.midi\n","\n","037.midi\n","\n","038.midi\n","\n","039.midi\n","\n","040.midi\n","\n","041.midi\n","\n","042.midi\n","\n","043.midi\n","\n","044.midi\n","\n","045.midi\n","\n","046.midi\n","\n","047.midi\n","\n","048.midi\n","\n","049.midi\n","\n","050.midi\n","\n","051.midi\n","\n","052.midi\n","\n","053.midi\n","\n","054.midi\n","\n","055.midi\n","\n","056.midi\n","\n","057.midi\n","\n","058.midi\n","\n","059.midi\n","\n","060.midi\n","\n","061.midi\n","\n","062.midi\n","\n","063.midi\n","\n","064.midi\n","\n","065.midi\n","\n","066.midi\n","\n","067.midi\n","\n","068.midi\n","\n","069.midi\n","\n","070.midi\n","\n","071.midi\n","\n","072.midi\n","\n","073.midi\n","\n","074.midi\n","\n","075.midi\n","\n","076.midi\n","\n","077.midi\n","\n","078.midi\n","\n","079.midi\n","\n","080.midi\n","\n","081.midi\n","\n","082.midi\n","\n","083.midi\n","\n","084.midi\n","\n","085.midi\n","\n","086.midi\n","\n","087.midi\n","\n","088.midi\n","\n","089.midi\n","\n","090.midi\n","\n","091.midi\n","\n","092.midi\n","\n","093.midi\n","\n","094.midi\n","\n","095.midi\n","\n","096.midi\n","\n","097.midi\n","\n","098.midi\n","\n","099.midi\n","\n","100.midi\n","\n","101.midi\n","\n","102.midi\n","\n","103.midi\n","\n","104.midi\n","\n","105.midi\n","\n","106.midi\n","\n","107.midi\n","\n","108.midi\n","\n","109.midi\n","\n","110.midi\n","\n","111.midi\n","\n","112.midi\n","\n","113.midi\n","\n","114.midi\n","\n","115.midi\n","\n","116.midi\n","\n","117.midi\n","\n","118.midi\n","\n","119.midi\n","\n","120.midi\n","\n","121.midi\n","\n","122.midi\n","\n","123.midi\n","\n","124.midi\n","\n","125.midi\n","\n","126.midi\n","\n","127.midi\n","\n","128.midi\n","\n","129.midi\n","\n","130.midi\n","\n","131.midi\n","\n","132.midi\n","\n","133.midi\n","\n","134.midi\n","\n","135.midi\n","\n","136.midi\n","\n","137.midi\n","\n","138.midi\n","\n","139.midi\n","\n","140.midi\n","\n","141.midi\n","\n","142.midi\n","\n","143.midi\n","\n","144.midi\n","\n","145.midi\n","\n","146.midi\n","\n","147.midi\n","\n","148.midi\n","\n","149.midi\n","\n","150.midi\n","\n","151.midi\n","\n","152.midi\n","\n","153.midi\n","\n","154.midi\n","\n","155.midi\n","\n","156.midi\n","\n","157.midi\n","\n","158.midi\n","\n","159.midi\n","\n","160.midi\n","\n","161.midi\n","\n","162.midi\n","\n","163.midi\n","\n","164.midi\n","\n","165.midi\n","\n","166.midi\n","\n","167.midi\n","\n","168.midi\n","\n","169.midi\n","\n","170.midi\n","\n","171.midi\n","\n","172.midi\n","\n","173.midi\n","\n","174.midi\n","\n","175.midi\n","\n","176.midi\n","\n","177.midi\n","\n","178.midi\n","\n","179.midi\n","\n","180.midi\n","\n","181.midi\n","\n","182.midi\n","\n","183.midi\n","\n","184.midi\n","\n","185.midi\n","\n","186.midi\n","\n","187.midi\n","\n","188.midi\n","\n","189.midi\n","\n","190.midi\n","\n","191.midi\n","\n","192.midi\n","\n","193.midi\n","\n","194.midi\n","\n","195.midi\n","\n","196.midi\n","\n","197.midi\n","\n","198.midi\n","\n","199.midi\n","\n","200.midi\n","\n","201.midi\n","\n","202.midi\n","\n","203.midi\n","\n","204.midi\n","\n","205.midi\n","\n","206.midi\n","\n","207.midi\n","\n","208.midi\n","\n","209.midi\n","\n","210.midi\n","\n","211.midi\n","\n","212.midi\n","\n","213.midi\n","\n","214.midi\n","\n","215.midi\n","\n","216.midi\n","\n","217.midi\n","\n","218.midi\n","\n","219.midi\n","\n","220.midi\n","\n","221.midi\n","\n","222.midi\n","\n","223.midi\n","\n","224.midi\n","\n","225.midi\n","\n","226.midi\n","\n","227.midi\n","\n","228.midi\n","\n","229.midi\n","\n","230.midi\n","\n","231.midi\n","\n","232.midi\n","\n","233.midi\n","\n","234.midi\n","\n","235.midi\n","\n","236.midi\n","\n","237.midi\n","\n","238.midi\n","\n","239.midi\n","\n","240.midi\n","\n","241.midi\n","\n","242.midi\n","\n","243.midi\n","\n","244.midi\n","\n","245.midi\n","\n","246.midi\n","\n","247.midi\n","\n","248.midi\n","\n","249.midi\n","\n","250.midi\n","\n","251.midi\n","\n","252.midi\n","\n","253.midi\n","\n","254.midi\n","\n","255.midi\n","\n","256.midi\n","\n","257.midi\n","\n","258.midi\n","\n","259.midi\n","\n","260.midi\n","\n","261.midi\n","\n","262.midi\n","\n","263.midi\n","\n","264.midi\n","\n","265.midi\n","\n","266.midi\n","\n","267.midi\n","\n","268.midi\n","\n","269.midi\n","\n","270.midi\n","\n","271.midi\n","\n","272.midi\n","\n","273.midi\n","\n","274.midi\n","\n","275.midi\n","\n","276.midi\n","\n","277.midi\n","\n","278.midi\n","\n","279.midi\n","\n","280.midi\n","\n","281.midi\n","\n","282.midi\n","\n","283.midi\n","\n","284.midi\n","\n","285.midi\n","\n","286.midi\n","\n","287.midi\n","\n","288.midi\n","\n","289.midi\n","\n","290.midi\n","\n","291.midi\n","\n","292.midi\n","\n","293.midi\n","\n","294.midi\n","\n","295.midi\n","\n","296.midi\n","\n","297.midi\n","\n","298.midi\n","\n","299.midi\n","\n","300.midi\n","\n","301.midi\n","\n","302.midi\n","\n","303.midi\n","\n","304.midi\n","\n","305.midi\n","\n","306.midi\n","\n","307.midi\n","\n","308.midi\n","\n","309.midi\n","\n","310.midi\n","\n","311.midi\n","\n","312.midi\n","\n","313.midi\n","\n","314.midi\n","\n","315.midi\n","\n","316.midi\n","\n","318.midi\n","\n","319.midi\n","\n","320.midi\n","\n","321.midi\n","\n","322.midi\n","\n","323.midi\n","\n","324.midi\n","\n","325.midi\n","\n","326.midi\n","\n","327.midi\n","\n","328.midi\n","\n","329.midi\n","\n","330.midi\n","\n","331.midi\n","\n","332.midi\n","\n","333.midi\n","\n","334.midi\n","\n","335.midi\n","\n","336.midi\n","\n","337.midi\n","\n","338.midi\n","\n","339.midi\n","\n","340.midi\n","\n","341.midi\n","\n","342.midi\n","\n","343.midi\n","\n","344.midi\n","\n","345.midi\n","\n","346.midi\n","\n","347.midi\n","\n","348.midi\n","\n","349.midi\n","\n","350.midi\n","\n","351.midi\n","\n","352.midi\n","\n","353.midi\n","\n","354.midi\n","\n","355.midi\n","\n","356.midi\n","\n","357.midi\n","\n","358.midi\n","\n","359.midi\n","\n","360.midi\n","\n","361.midi\n","\n","362.midi\n","\n","363.midi\n","\n","364.midi\n","\n","365.midi\n","\n","366.midi\n","\n","367.midi\n","\n","368.midi\n","\n","369.midi\n","\n","370.midi\n","\n","371.midi\n","\n","372.midi\n","\n","373.midi\n","\n","374.midi\n","\n","375.midi\n","\n","376.midi\n","\n","377.midi\n","\n","378.midi\n","\n","379.midi\n","\n","380.midi\n","\n","381.midi\n","\n","382.midi\n","\n","383.midi\n","\n","384.midi\n","\n","385.midi\n","\n","386.midi\n","\n","387.midi\n","\n","388.midi\n","\n","389.midi\n","\n","390.midi\n","\n","391.midi\n","\n","392.midi\n","\n","393.midi\n","\n","394.midi\n","\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"t9eaGFHoyUkV","colab_type":"text"},"source":["Analyzing data"]},{"cell_type":"code","metadata":{"id":"WiOPUP7Oxlaq","colab_type":"code","outputId":"1145d403-5ce3-4f80-bb66-dd5c792e33b6","executionInfo":{"status":"ok","timestamp":1587370391959,"user_tz":240,"elapsed":88469,"user":{"displayName":"Sadman Sakif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhFzTKph8jrvXjeSmjPv-AWjlCXPOb8wSzBHYRB7w=s64","userId":"15126335579010581692"}},"colab":{"base_uri":"https://localhost:8080/","height":400}},"source":["unique_notes = list(set(notes))\n","freq = dict(Counter(notes))\n","\n","#consider only the frequencies\n","no=[count for _,count in freq.items()]\n","\n","#set the figure size\n","plt.figure(figsize=(15,5))\n","\n","#plot\n","plt.hist(no)"],"execution_count":57,"outputs":[{"output_type":"execute_result","data":{"text/plain":["(array([352.,  21.,   6.,   3.,   5.,   5.,   3.,   3.,   1.,   2.]),\n"," array([1.0000e+00, 1.0430e+03, 2.0850e+03, 3.1270e+03, 4.1690e+03,\n","        5.2110e+03, 6.2530e+03, 7.2950e+03, 8.3370e+03, 9.3790e+03,\n","        1.0421e+04]),\n"," <a list of 10 Patch objects>)"]},"metadata":{"tags":[]},"execution_count":57},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 1080x360 with 1 Axes>"]},"metadata":{"tags":[],"image/png":{"width":877,"height":302},"needs_background":"light"}}]},{"cell_type":"code","metadata":{"id":"KrJWm9ltzMIK","colab_type":"code","outputId":"d550fee8-ceff-45ca-e778-b48b8acd7afc","executionInfo":{"status":"ok","timestamp":1587370391960,"user_tz":240,"elapsed":84525,"user":{"displayName":"Sadman Sakif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhFzTKph8jrvXjeSmjPv-AWjlCXPOb8wSzBHYRB7w=s64","userId":"15126335579010581692"}},"colab":{"base_uri":"https://localhost:8080/","height":32}},"source":["frequent_notes = [notes for notes, count in freq.items() if count>=10]\n","print(len(frequent_notes))"],"execution_count":58,"outputs":[{"output_type":"stream","text":["272\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"dF3IgrqX7cHG","colab_type":"code","outputId":"6a0b1361-d6eb-498f-c46b-f973d8a33d1f","executionInfo":{"status":"ok","timestamp":1587370391961,"user_tz":240,"elapsed":82768,"user":{"displayName":"Sadman Sakif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhFzTKph8jrvXjeSmjPv-AWjlCXPOb8wSzBHYRB7w=s64","userId":"15126335579010581692"}},"colab":{"base_uri":"https://localhost:8080/","height":32}},"source":["len(unique_notes)"],"execution_count":59,"outputs":[{"output_type":"execute_result","data":{"text/plain":["401"]},"metadata":{"tags":[]},"execution_count":59}]},{"cell_type":"markdown","metadata":{"id":"fjEUiFD3LpVF","colab_type":"text"},"source":["New array that contains only the top frequency notes"]},{"cell_type":"code","metadata":{"id":"vAp3IEaaLxPt","colab_type":"code","colab":{}},"source":["new_music = []\n","for elm in all_notes:\n","  temp = []\n","  for a in elm:\n","    if a in frequent_notes:\n","      temp.append(a)\n","  new_music.append(temp)\n","\n","new_music = np.array(new_music)"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"84HYz4NShp8c","colab_type":"code","colab":{}},"source":["new_music.shape"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"7Dh1YNSwd7WR","colab_type":"text"},"source":["Preparing input and output sequences"]},{"cell_type":"code","metadata":{"id":"ngAk-oyPdMAp","colab_type":"code","colab":{}},"source":["#length of a input sequence\n","no_of_timesteps = 32\n","\n","x = []\n","y = []\n","\n","for elm in new_music:\n","  for i in range(0, len(elm) - no_of_timesteps, 1):\n","\n","    #preparing input and output sequences\n","    inp = elm[i:i + no_of_timesteps]\n","    out = elm[i + no_of_timesteps]\n","\n","    x.append(inp)\n","    y.append(out)\n","\n","x = np.array(x)\n","y = np.array(y)\n","\n","\n","#no. of unique notes\n","#n_vocab = len(set(notes))  \n","#all the unique notes\n","#pitch = sorted(set(item for item in notes))  \n","#assign unique value to every note\n","#note_to_int = dict((note, number) for number, note in enumerate(pitch))\n","#preparing input and output sequences\n","#X = []\n","#y = []\n","#for notes in all_notes:\n","#  for i in range(0, len(notes) - no_of_timesteps, 1):\n","#    input_ = notes[i:i + no_of_timesteps]\n","#    output = notes[i + no_of_timesteps]\n","#    X.append([note_to_int[note] for note in input_])\n","#    y.append(note_to_int[output])"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"1WbFCtwdh4o7","colab_type":"code","colab":{}},"source":["x[1]"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"MWJACnmgldIe","colab_type":"code","colab":{}},"source":["y[1]"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"kioE4ryLPzpu","colab_type":"text"},"source":["Assigning a unique integer to every input note:"]},{"cell_type":"code","metadata":{"id":"5sENcLWFc06E","colab_type":"code","colab":{}},"source":["unique_x = list(set(x.ravel()))\n","x_note_to_int = dict((note_, number) for number, note_ in enumerate(unique_x))\n","\n","x_seq = []\n","for i in x:\n","  temp = []\n","  for j in i:\n","    temp.append(x_note_to_int[j])\n","  x_seq.append(temp)\n","\n","x_seq = np.array(x_seq)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"V_oI1XbmSESd","colab_type":"text"},"source":["Assigning a unique integer to every output note:"]},{"cell_type":"code","metadata":{"id":"ZAnAar4ISH5h","colab_type":"code","colab":{}},"source":["unique_y = list(set(y))\n","y_note_to_int = dict((note_, number) for number, note_ in enumerate(unique_y)) \n","y_seq=np.array([y_note_to_int[i] for i in y])"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"rQtFDV3jSWx6","colab_type":"text"},"source":["80% of the data for training and the rest 20% for the validation:"]},{"cell_type":"code","metadata":{"id":"oWGamdAOSbUw","colab_type":"code","colab":{}},"source":["x_tr, x_val, y_tr, y_val = train_test_split(x_seq,y_seq,test_size=0.2,random_state=0)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"UJ-lwcGVfRpW","colab_type":"text"},"source":["Reshaping the input for the Convolution 1D and then normalizing"]},{"cell_type":"code","metadata":{"id":"EfO3gQdvePz0","colab_type":"code","colab":{}},"source":["#reshaping\n","#X = np.reshape(X, (len(X), no_of_timesteps, 1))\n","#normalizing the inputs\n","#X = X / float(n_vocab) "],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"-z1wq6akg6mw","colab_type":"text"},"source":["WaveNet Model"]},{"cell_type":"code","metadata":{"id":"RmTU3zHqU5r0","colab_type":"code","outputId":"03eab6ac-edc3-4a15-a7f0-02dd8eca5125","executionInfo":{"status":"ok","timestamp":1587370401580,"user_tz":240,"elapsed":49163,"user":{"displayName":"Sadman Sakif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhFzTKph8jrvXjeSmjPv-AWjlCXPOb8wSzBHYRB7w=s64","userId":"15126335579010581692"}},"colab":{"base_uri":"https://localhost:8080/","height":576}},"source":["K.clear_session()\n","model = Sequential()\n","    \n","#embedding layer\n","model.add(Embedding(len(unique_x), 100, input_length=32,trainable=True)) \n","\n","model.add(Conv1D(64,3, padding='causal',activation='relu'))\n","model.add(Dropout(0.2))\n","model.add(MaxPool1D(2))\n","    \n","model.add(Conv1D(128,3,activation='relu',dilation_rate=2,padding='causal'))\n","model.add(Dropout(0.2))\n","model.add(MaxPool1D(2))\n","\n","model.add(Conv1D(256,3,activation='relu',dilation_rate=4,padding='causal'))\n","model.add(Dropout(0.2))\n","model.add(MaxPool1D(2))\n","          \n","#model.add(Conv1D(256,5,activation='relu'))    \n","model.add(GlobalMaxPool1D())\n","    \n","model.add(Dense(256, activation='relu'))\n","model.add(Dense(len(unique_y), activation='softmax'))\n","    \n","model.compile(loss='sparse_categorical_crossentropy', optimizer='adam')\n","\n","model.summary()"],"execution_count":64,"outputs":[{"output_type":"stream","text":["Model: \"sequential_1\"\n","_________________________________________________________________\n","Layer (type)                 Output Shape              Param #   \n","=================================================================\n","embedding_1 (Embedding)      (None, 32, 100)           27200     \n","_________________________________________________________________\n","conv1d_1 (Conv1D)            (None, 32, 64)            19264     \n","_________________________________________________________________\n","dropout_1 (Dropout)          (None, 32, 64)            0         \n","_________________________________________________________________\n","max_pooling1d_1 (MaxPooling1 (None, 16, 64)            0         \n","_________________________________________________________________\n","conv1d_2 (Conv1D)            (None, 16, 128)           24704     \n","_________________________________________________________________\n","dropout_2 (Dropout)          (None, 16, 128)           0         \n","_________________________________________________________________\n","max_pooling1d_2 (MaxPooling1 (None, 8, 128)            0         \n","_________________________________________________________________\n","conv1d_3 (Conv1D)            (None, 8, 256)            98560     \n","_________________________________________________________________\n","dropout_3 (Dropout)          (None, 8, 256)            0         \n","_________________________________________________________________\n","max_pooling1d_3 (MaxPooling1 (None, 4, 256)            0         \n","_________________________________________________________________\n","global_max_pooling1d_1 (Glob (None, 256)               0         \n","_________________________________________________________________\n","dense_1 (Dense)              (None, 256)               65792     \n","_________________________________________________________________\n","dense_2 (Dense)              (None, 272)               69904     \n","=================================================================\n","Total params: 305,424\n","Trainable params: 305,424\n","Non-trainable params: 0\n","_________________________________________________________________\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"D-gcZU4bBw9U","colab_type":"code","colab":{}},"source":["#K.clear_session()\n","#def simple_wavenet():\n","#  no_of_kernels=64\n","#  num_of_blocks= int(np.sqrt(no_of_timesteps)) - 1   #no. of stacked conv1d layers\n","\n","#  model = Sequential()\n","#  for i in range(num_of_blocks):\n","#    model.add(Conv1D(no_of_kernels,3,dilation_rate=(2**i),padding='causal',activation='relu'))\n","#  model.add(Conv1D(1, 1, activation='relu', padding='causal'))\n","#  model.add(Flatten())\n","#  model.add(Dense(128, activation='relu'))\n","#  model.add(Dense(n_vocab, activation='softmax'))\n","#  model.compile(loss='sparse_categorical_crossentropy', optimizer='adam')\n","#  return model"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"dzofSxuzV6Z1","colab_type":"text"},"source":["Save the best model during training"]},{"cell_type":"code","metadata":{"id":"EXKVjhQiV77S","colab_type":"code","colab":{}},"source":["mc=ModelCheckpoint('drive/My Drive/ECE496 CNN/Saved Models/best_model.h5', monitor='val_loss', mode='min', save_best_only=True,verbose=1)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"pi93GqD1FFDx"},"source":["\n","Save the model after every 10 epochs"]},{"cell_type":"code","metadata":{"id":"97xRIhZcUGGn","colab_type":"code","colab":{}},"source":["#mc = keras.callbacks.ModelCheckpoint('drive/My Drive/ECE496 CNN/Saved Models/model{epoch:03d}.h5', save_weights_only=False, period=10)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"qkN1YAy5WEOd","colab_type":"text"},"source":["Train the model with a batch size of 128 for 50 epochs"]},{"cell_type":"code","metadata":{"id":"FHYxSHKnWFgA","colab_type":"code","outputId":"0d9e5afa-9e67-4ffd-ad0e-8d26ad6f1db0","executionInfo":{"status":"ok","timestamp":1587366879074,"user_tz":240,"elapsed":2221486,"user":{"displayName":"Sadman Sakif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhFzTKph8jrvXjeSmjPv-AWjlCXPOb8wSzBHYRB7w=s64","userId":"15126335579010581692"}},"colab":{"base_uri":"https://localhost:8080/","height":1000}},"source":["history = model.fit(np.array(x_tr),np.array(y_tr),batch_size=128,epochs=200, validation_data=(np.array(x_val),np.array(y_val)),verbose=1, callbacks=[mc])"],"execution_count":0,"outputs":[{"output_type":"stream","text":["WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:422: The name tf.global_variables is deprecated. Please use tf.compat.v1.global_variables instead.\n","\n","Train on 179364 samples, validate on 44841 samples\n","Epoch 1/200\n","179364/179364 [==============================] - 17s 97us/step - loss: 3.4919 - val_loss: 3.3255\n","\n","Epoch 00001: val_loss improved from inf to 3.32554, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 2/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 3.0849 - val_loss: 3.2023\n","\n","Epoch 00002: val_loss improved from 3.32554 to 3.20227, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 3/200\n","179364/179364 [==============================] - 12s 65us/step - loss: 2.9618 - val_loss: 3.0704\n","\n","Epoch 00003: val_loss improved from 3.20227 to 3.07036, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 4/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.8867 - val_loss: 3.0137\n","\n","Epoch 00004: val_loss improved from 3.07036 to 3.01369, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 5/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.8310 - val_loss: 2.9066\n","\n","Epoch 00005: val_loss improved from 3.01369 to 2.90662, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 6/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.7851 - val_loss: 2.9140\n","\n","Epoch 00006: val_loss did not improve from 2.90662\n","Epoch 7/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.7517 - val_loss: 2.8825\n","\n","Epoch 00007: val_loss improved from 2.90662 to 2.88247, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 8/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.7227 - val_loss: 2.8387\n","\n","Epoch 00008: val_loss improved from 2.88247 to 2.83875, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 9/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.6959 - val_loss: 2.8040\n","\n","Epoch 00009: val_loss improved from 2.83875 to 2.80399, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 10/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.6729 - val_loss: 2.7914\n","\n","Epoch 00010: val_loss improved from 2.80399 to 2.79139, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 11/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.6519 - val_loss: 2.7576\n","\n","Epoch 00011: val_loss improved from 2.79139 to 2.75756, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 12/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.6360 - val_loss: 2.7671\n","\n","Epoch 00012: val_loss did not improve from 2.75756\n","Epoch 13/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.6210 - val_loss: 2.7509\n","\n","Epoch 00013: val_loss improved from 2.75756 to 2.75088, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 14/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.6100 - val_loss: 2.7364\n","\n","Epoch 00014: val_loss improved from 2.75088 to 2.73639, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 15/200\n","179364/179364 [==============================] - 11s 64us/step - loss: 2.5991 - val_loss: 2.7170\n","\n","Epoch 00015: val_loss improved from 2.73639 to 2.71699, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 16/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.5872 - val_loss: 2.7353\n","\n","Epoch 00016: val_loss did not improve from 2.71699\n","Epoch 17/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.5789 - val_loss: 2.7058\n","\n","Epoch 00017: val_loss improved from 2.71699 to 2.70580, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 18/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.5698 - val_loss: 2.7133\n","\n","Epoch 00018: val_loss did not improve from 2.70580\n","Epoch 19/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.5620 - val_loss: 2.7152\n","\n","Epoch 00019: val_loss did not improve from 2.70580\n","Epoch 20/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.5532 - val_loss: 2.6935\n","\n","Epoch 00020: val_loss improved from 2.70580 to 2.69350, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 21/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.5455 - val_loss: 2.6812\n","\n","Epoch 00021: val_loss improved from 2.69350 to 2.68122, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 22/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.5379 - val_loss: 2.6753\n","\n","Epoch 00022: val_loss improved from 2.68122 to 2.67526, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 23/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.5362 - val_loss: 2.6668\n","\n","Epoch 00023: val_loss improved from 2.67526 to 2.66682, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 24/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.5309 - val_loss: 2.6727\n","\n","Epoch 00024: val_loss did not improve from 2.66682\n","Epoch 25/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.5237 - val_loss: 2.6621\n","\n","Epoch 00025: val_loss improved from 2.66682 to 2.66207, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 26/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.5189 - val_loss: 2.6530\n","\n","Epoch 00026: val_loss improved from 2.66207 to 2.65300, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 27/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.5158 - val_loss: 2.6637\n","\n","Epoch 00027: val_loss did not improve from 2.65300\n","Epoch 28/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.5110 - val_loss: 2.6531\n","\n","Epoch 00028: val_loss did not improve from 2.65300\n","Epoch 29/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.5064 - val_loss: 2.6591\n","\n","Epoch 00029: val_loss did not improve from 2.65300\n","Epoch 30/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.5052 - val_loss: 2.6436\n","\n","Epoch 00030: val_loss improved from 2.65300 to 2.64356, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 31/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.5010 - val_loss: 2.6557\n","\n","Epoch 00031: val_loss did not improve from 2.64356\n","Epoch 32/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4955 - val_loss: 2.6423\n","\n","Epoch 00032: val_loss improved from 2.64356 to 2.64234, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 33/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4936 - val_loss: 2.6501\n","\n","Epoch 00033: val_loss did not improve from 2.64234\n","Epoch 34/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4902 - val_loss: 2.6455\n","\n","Epoch 00034: val_loss did not improve from 2.64234\n","Epoch 35/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4911 - val_loss: 2.6487\n","\n","Epoch 00035: val_loss did not improve from 2.64234\n","Epoch 36/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4854 - val_loss: 2.6393\n","\n","Epoch 00036: val_loss improved from 2.64234 to 2.63931, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 37/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4869 - val_loss: 2.6421\n","\n","Epoch 00037: val_loss did not improve from 2.63931\n","Epoch 38/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4824 - val_loss: 2.6467\n","\n","Epoch 00038: val_loss did not improve from 2.63931\n","Epoch 39/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4795 - val_loss: 2.6364\n","\n","Epoch 00039: val_loss improved from 2.63931 to 2.63635, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 40/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4758 - val_loss: 2.6280\n","\n","Epoch 00040: val_loss improved from 2.63635 to 2.62796, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 41/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4713 - val_loss: 2.6417\n","\n","Epoch 00041: val_loss did not improve from 2.62796\n","Epoch 42/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4707 - val_loss: 2.6234\n","\n","Epoch 00042: val_loss improved from 2.62796 to 2.62335, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 43/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4693 - val_loss: 2.6390\n","\n","Epoch 00043: val_loss did not improve from 2.62335\n","Epoch 44/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4696 - val_loss: 2.6246\n","\n","Epoch 00044: val_loss did not improve from 2.62335\n","Epoch 45/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4632 - val_loss: 2.6279\n","\n","Epoch 00045: val_loss did not improve from 2.62335\n","Epoch 46/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4653 - val_loss: 2.6229\n","\n","Epoch 00046: val_loss improved from 2.62335 to 2.62292, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 47/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4641 - val_loss: 2.6301\n","\n","Epoch 00047: val_loss did not improve from 2.62292\n","Epoch 48/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4577 - val_loss: 2.6408\n","\n","Epoch 00048: val_loss did not improve from 2.62292\n","Epoch 49/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4563 - val_loss: 2.6201\n","\n","Epoch 00049: val_loss improved from 2.62292 to 2.62013, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 50/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4543 - val_loss: 2.6261\n","\n","Epoch 00050: val_loss did not improve from 2.62013\n","Epoch 51/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4536 - val_loss: 2.6302\n","\n","Epoch 00051: val_loss did not improve from 2.62013\n","Epoch 52/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4539 - val_loss: 2.6203\n","\n","Epoch 00052: val_loss did not improve from 2.62013\n","Epoch 53/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4515 - val_loss: 2.6185\n","\n","Epoch 00053: val_loss improved from 2.62013 to 2.61851, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 54/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4478 - val_loss: 2.6280\n","\n","Epoch 00054: val_loss did not improve from 2.61851\n","Epoch 55/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4531 - val_loss: 2.6257\n","\n","Epoch 00055: val_loss did not improve from 2.61851\n","Epoch 56/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4499 - val_loss: 2.6353\n","\n","Epoch 00056: val_loss did not improve from 2.61851\n","Epoch 57/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4495 - val_loss: 2.6351\n","\n","Epoch 00057: val_loss did not improve from 2.61851\n","Epoch 58/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4410 - val_loss: 2.6161\n","\n","Epoch 00058: val_loss improved from 2.61851 to 2.61606, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 59/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4450 - val_loss: 2.6181\n","\n","Epoch 00059: val_loss did not improve from 2.61606\n","Epoch 60/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4423 - val_loss: 2.6228\n","\n","Epoch 00060: val_loss did not improve from 2.61606\n","Epoch 61/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4460 - val_loss: 2.6171\n","\n","Epoch 00061: val_loss did not improve from 2.61606\n","Epoch 62/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4393 - val_loss: 2.6181\n","\n","Epoch 00062: val_loss did not improve from 2.61606\n","Epoch 63/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4390 - val_loss: 2.6140\n","\n","Epoch 00063: val_loss improved from 2.61606 to 2.61400, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 64/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4376 - val_loss: 2.6143\n","\n","Epoch 00064: val_loss did not improve from 2.61400\n","Epoch 65/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4390 - val_loss: 2.6198\n","\n","Epoch 00065: val_loss did not improve from 2.61400\n","Epoch 66/200\n","179364/179364 [==============================] - 12s 64us/step - loss: 2.4353 - val_loss: 2.6133\n","\n","Epoch 00066: val_loss improved from 2.61400 to 2.61331, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 67/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4336 - val_loss: 2.6224\n","\n","Epoch 00067: val_loss did not improve from 2.61331\n","Epoch 68/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4344 - val_loss: 2.6192\n","\n","Epoch 00068: val_loss did not improve from 2.61331\n","Epoch 69/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4336 - val_loss: 2.6293\n","\n","Epoch 00069: val_loss did not improve from 2.61331\n","Epoch 70/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4339 - val_loss: 2.6165\n","\n","Epoch 00070: val_loss did not improve from 2.61331\n","Epoch 71/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4301 - val_loss: 2.6158\n","\n","Epoch 00071: val_loss did not improve from 2.61331\n","Epoch 72/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4345 - val_loss: 2.6217\n","\n","Epoch 00072: val_loss did not improve from 2.61331\n","Epoch 73/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4320 - val_loss: 2.6254\n","\n","Epoch 00073: val_loss did not improve from 2.61331\n","Epoch 74/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4293 - val_loss: 2.6196\n","\n","Epoch 00074: val_loss did not improve from 2.61331\n","Epoch 75/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4303 - val_loss: 2.6006\n","\n","Epoch 00075: val_loss improved from 2.61331 to 2.60055, saving model to drive/My Drive/ECE496 CNN/Saved Models/best_model.h5\n","Epoch 76/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4282 - val_loss: 2.6181\n","\n","Epoch 00076: val_loss did not improve from 2.60055\n","Epoch 77/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4273 - val_loss: 2.6128\n","\n","Epoch 00077: val_loss did not improve from 2.60055\n","Epoch 78/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4274 - val_loss: 2.6192\n","\n","Epoch 00078: val_loss did not improve from 2.60055\n","Epoch 79/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4230 - val_loss: 2.6168\n","\n","Epoch 00079: val_loss did not improve from 2.60055\n","Epoch 80/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4246 - val_loss: 2.6151\n","\n","Epoch 00080: val_loss did not improve from 2.60055\n","Epoch 81/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4254 - val_loss: 2.6224\n","\n","Epoch 00081: val_loss did not improve from 2.60055\n","Epoch 82/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4185 - val_loss: 2.6131\n","\n","Epoch 00082: val_loss did not improve from 2.60055\n","Epoch 83/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4243 - val_loss: 2.6149\n","\n","Epoch 00083: val_loss did not improve from 2.60055\n","Epoch 84/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4233 - val_loss: 2.6211\n","\n","Epoch 00084: val_loss did not improve from 2.60055\n","Epoch 85/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4219 - val_loss: 2.6177\n","\n","Epoch 00085: val_loss did not improve from 2.60055\n","Epoch 86/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4186 - val_loss: 2.6125\n","\n","Epoch 00086: val_loss did not improve from 2.60055\n","Epoch 87/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4191 - val_loss: 2.6270\n","\n","Epoch 00087: val_loss did not improve from 2.60055\n","Epoch 88/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4203 - val_loss: 2.6260\n","\n","Epoch 00088: val_loss did not improve from 2.60055\n","Epoch 89/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4220 - val_loss: 2.6179\n","\n","Epoch 00089: val_loss did not improve from 2.60055\n","Epoch 90/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4173 - val_loss: 2.6141\n","\n","Epoch 00090: val_loss did not improve from 2.60055\n","Epoch 91/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4219 - val_loss: 2.6184\n","\n","Epoch 00091: val_loss did not improve from 2.60055\n","Epoch 92/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4168 - val_loss: 2.6093\n","\n","Epoch 00092: val_loss did not improve from 2.60055\n","Epoch 93/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4193 - val_loss: 2.6255\n","\n","Epoch 00093: val_loss did not improve from 2.60055\n","Epoch 94/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4174 - val_loss: 2.6249\n","\n","Epoch 00094: val_loss did not improve from 2.60055\n","Epoch 95/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4158 - val_loss: 2.6134\n","\n","Epoch 00095: val_loss did not improve from 2.60055\n","Epoch 96/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4168 - val_loss: 2.6187\n","\n","Epoch 00096: val_loss did not improve from 2.60055\n","Epoch 97/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4149 - val_loss: 2.6241\n","\n","Epoch 00097: val_loss did not improve from 2.60055\n","Epoch 98/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4159 - val_loss: 2.6256\n","\n","Epoch 00098: val_loss did not improve from 2.60055\n","Epoch 99/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4166 - val_loss: 2.6198\n","\n","Epoch 00099: val_loss did not improve from 2.60055\n","Epoch 100/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4140 - val_loss: 2.6285\n","\n","Epoch 00100: val_loss did not improve from 2.60055\n","Epoch 101/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4170 - val_loss: 2.6160\n","\n","Epoch 00101: val_loss did not improve from 2.60055\n","Epoch 102/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4145 - val_loss: 2.6317\n","\n","Epoch 00102: val_loss did not improve from 2.60055\n","Epoch 103/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4148 - val_loss: 2.6187\n","\n","Epoch 00103: val_loss did not improve from 2.60055\n","Epoch 104/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4121 - val_loss: 2.6191\n","\n","Epoch 00104: val_loss did not improve from 2.60055\n","Epoch 105/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4135 - val_loss: 2.6211\n","\n","Epoch 00105: val_loss did not improve from 2.60055\n","Epoch 106/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4156 - val_loss: 2.6102\n","\n","Epoch 00106: val_loss did not improve from 2.60055\n","Epoch 107/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4131 - val_loss: 2.6063\n","\n","Epoch 00107: val_loss did not improve from 2.60055\n","Epoch 108/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4155 - val_loss: 2.6150\n","\n","Epoch 00108: val_loss did not improve from 2.60055\n","Epoch 109/200\n","179364/179364 [==============================] - 10s 58us/step - loss: 2.4122 - val_loss: 2.6120\n","\n","Epoch 00109: val_loss did not improve from 2.60055\n","Epoch 110/200\n","179364/179364 [==============================] - 10s 58us/step - loss: 2.4101 - val_loss: 2.6208\n","\n","Epoch 00110: val_loss did not improve from 2.60055\n","Epoch 111/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4127 - val_loss: 2.6206\n","\n","Epoch 00111: val_loss did not improve from 2.60055\n","Epoch 112/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4123 - val_loss: 2.6069\n","\n","Epoch 00112: val_loss did not improve from 2.60055\n","Epoch 113/200\n","179364/179364 [==============================] - 10s 57us/step - loss: 2.4098 - val_loss: 2.6261\n","\n","Epoch 00113: val_loss did not improve from 2.60055\n","Epoch 114/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4127 - val_loss: 2.6180\n","\n","Epoch 00114: val_loss did not improve from 2.60055\n","Epoch 115/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4109 - val_loss: 2.6258\n","\n","Epoch 00115: val_loss did not improve from 2.60055\n","Epoch 116/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4091 - val_loss: 2.6179\n","\n","Epoch 00116: val_loss did not improve from 2.60055\n","Epoch 117/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4087 - val_loss: 2.6267\n","\n","Epoch 00117: val_loss did not improve from 2.60055\n","Epoch 118/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4100 - val_loss: 2.6217\n","\n","Epoch 00118: val_loss did not improve from 2.60055\n","Epoch 119/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4104 - val_loss: 2.6165\n","\n","Epoch 00119: val_loss did not improve from 2.60055\n","Epoch 120/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4067 - val_loss: 2.6141\n","\n","Epoch 00120: val_loss did not improve from 2.60055\n","Epoch 121/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4067 - val_loss: 2.6203\n","\n","Epoch 00121: val_loss did not improve from 2.60055\n","Epoch 122/200\n","179364/179364 [==============================] - 11s 64us/step - loss: 2.4097 - val_loss: 2.6178\n","\n","Epoch 00122: val_loss did not improve from 2.60055\n","Epoch 123/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4069 - val_loss: 2.6193\n","\n","Epoch 00123: val_loss did not improve from 2.60055\n","Epoch 124/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4044 - val_loss: 2.6193\n","\n","Epoch 00124: val_loss did not improve from 2.60055\n","Epoch 125/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4100 - val_loss: 2.6342\n","\n","Epoch 00125: val_loss did not improve from 2.60055\n","Epoch 126/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4056 - val_loss: 2.6280\n","\n","Epoch 00126: val_loss did not improve from 2.60055\n","Epoch 127/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4079 - val_loss: 2.6130\n","\n","Epoch 00127: val_loss did not improve from 2.60055\n","Epoch 128/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4062 - val_loss: 2.6187\n","\n","Epoch 00128: val_loss did not improve from 2.60055\n","Epoch 129/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4059 - val_loss: 2.6169\n","\n","Epoch 00129: val_loss did not improve from 2.60055\n","Epoch 130/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4093 - val_loss: 2.6197\n","\n","Epoch 00130: val_loss did not improve from 2.60055\n","Epoch 131/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4036 - val_loss: 2.6300\n","\n","Epoch 00131: val_loss did not improve from 2.60055\n","Epoch 132/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4051 - val_loss: 2.6244\n","\n","Epoch 00132: val_loss did not improve from 2.60055\n","Epoch 133/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4059 - val_loss: 2.6274\n","\n","Epoch 00133: val_loss did not improve from 2.60055\n","Epoch 134/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4065 - val_loss: 2.6235\n","\n","Epoch 00134: val_loss did not improve from 2.60055\n","Epoch 135/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4128 - val_loss: 2.6340\n","\n","Epoch 00135: val_loss did not improve from 2.60055\n","Epoch 136/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4071 - val_loss: 2.6283\n","\n","Epoch 00136: val_loss did not improve from 2.60055\n","Epoch 137/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4046 - val_loss: 2.6223\n","\n","Epoch 00137: val_loss did not improve from 2.60055\n","Epoch 138/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4069 - val_loss: 2.6216\n","\n","Epoch 00138: val_loss did not improve from 2.60055\n","Epoch 139/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4071 - val_loss: 2.6269\n","\n","Epoch 00139: val_loss did not improve from 2.60055\n","Epoch 140/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4016 - val_loss: 2.6181\n","\n","Epoch 00140: val_loss did not improve from 2.60055\n","Epoch 141/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4049 - val_loss: 2.6198\n","\n","Epoch 00141: val_loss did not improve from 2.60055\n","Epoch 142/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4043 - val_loss: 2.6215\n","\n","Epoch 00142: val_loss did not improve from 2.60055\n","Epoch 143/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4070 - val_loss: 2.6452\n","\n","Epoch 00143: val_loss did not improve from 2.60055\n","Epoch 144/200\n","179364/179364 [==============================] - 10s 58us/step - loss: 2.4061 - val_loss: 2.6156\n","\n","Epoch 00144: val_loss did not improve from 2.60055\n","Epoch 145/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4047 - val_loss: 2.6258\n","\n","Epoch 00145: val_loss did not improve from 2.60055\n","Epoch 146/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4040 - val_loss: 2.6258\n","\n","Epoch 00146: val_loss did not improve from 2.60055\n","Epoch 147/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4071 - val_loss: 2.6233\n","\n","Epoch 00147: val_loss did not improve from 2.60055\n","Epoch 148/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4029 - val_loss: 2.6350\n","\n","Epoch 00148: val_loss did not improve from 2.60055\n","Epoch 149/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4055 - val_loss: 2.6167\n","\n","Epoch 00149: val_loss did not improve from 2.60055\n","Epoch 150/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4012 - val_loss: 2.6199\n","\n","Epoch 00150: val_loss did not improve from 2.60055\n","Epoch 151/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4053 - val_loss: 2.6164\n","\n","Epoch 00151: val_loss did not improve from 2.60055\n","Epoch 152/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4041 - val_loss: 2.6239\n","\n","Epoch 00152: val_loss did not improve from 2.60055\n","Epoch 153/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4031 - val_loss: 2.6349\n","\n","Epoch 00153: val_loss did not improve from 2.60055\n","Epoch 154/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4035 - val_loss: 2.6303\n","\n","Epoch 00154: val_loss did not improve from 2.60055\n","Epoch 155/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4011 - val_loss: 2.6258\n","\n","Epoch 00155: val_loss did not improve from 2.60055\n","Epoch 156/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4018 - val_loss: 2.6271\n","\n","Epoch 00156: val_loss did not improve from 2.60055\n","Epoch 157/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4059 - val_loss: 2.6251\n","\n","Epoch 00157: val_loss did not improve from 2.60055\n","Epoch 158/200\n","179364/179364 [==============================] - 10s 58us/step - loss: 2.4070 - val_loss: 2.6226\n","\n","Epoch 00158: val_loss did not improve from 2.60055\n","Epoch 159/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4058 - val_loss: 2.6203\n","\n","Epoch 00159: val_loss did not improve from 2.60055\n","Epoch 160/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4004 - val_loss: 2.6335\n","\n","Epoch 00160: val_loss did not improve from 2.60055\n","Epoch 161/200\n","179364/179364 [==============================] - 11s 59us/step - loss: 2.4049 - val_loss: 2.6335\n","\n","Epoch 00161: val_loss did not improve from 2.60055\n","Epoch 162/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4042 - val_loss: 2.6327\n","\n","Epoch 00162: val_loss did not improve from 2.60055\n","Epoch 163/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4055 - val_loss: 2.6238\n","\n","Epoch 00163: val_loss did not improve from 2.60055\n","Epoch 164/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4035 - val_loss: 2.6217\n","\n","Epoch 00164: val_loss did not improve from 2.60055\n","Epoch 165/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4031 - val_loss: 2.6262\n","\n","Epoch 00165: val_loss did not improve from 2.60055\n","Epoch 166/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4059 - val_loss: 2.6319\n","\n","Epoch 00166: val_loss did not improve from 2.60055\n","Epoch 167/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4008 - val_loss: 2.6338\n","\n","Epoch 00167: val_loss did not improve from 2.60055\n","Epoch 168/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4041 - val_loss: 2.6259\n","\n","Epoch 00168: val_loss did not improve from 2.60055\n","Epoch 169/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4077 - val_loss: 2.6239\n","\n","Epoch 00169: val_loss did not improve from 2.60055\n","Epoch 170/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4020 - val_loss: 2.6325\n","\n","Epoch 00170: val_loss did not improve from 2.60055\n","Epoch 171/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4032 - val_loss: 2.6183\n","\n","Epoch 00171: val_loss did not improve from 2.60055\n","Epoch 172/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4076 - val_loss: 2.6294\n","\n","Epoch 00172: val_loss did not improve from 2.60055\n","Epoch 173/200\n","179364/179364 [==============================] - 11s 60us/step - loss: 2.4029 - val_loss: 2.6100\n","\n","Epoch 00173: val_loss did not improve from 2.60055\n","Epoch 174/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4035 - val_loss: 2.6334\n","\n","Epoch 00174: val_loss did not improve from 2.60055\n","Epoch 175/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4010 - val_loss: 2.6372\n","\n","Epoch 00175: val_loss did not improve from 2.60055\n","Epoch 176/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4038 - val_loss: 2.6312\n","\n","Epoch 00176: val_loss did not improve from 2.60055\n","Epoch 177/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4032 - val_loss: 2.6334\n","\n","Epoch 00177: val_loss did not improve from 2.60055\n","Epoch 178/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4035 - val_loss: 2.6390\n","\n","Epoch 00178: val_loss did not improve from 2.60055\n","Epoch 179/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4043 - val_loss: 2.6341\n","\n","Epoch 00179: val_loss did not improve from 2.60055\n","Epoch 180/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4092 - val_loss: 2.6259\n","\n","Epoch 00180: val_loss did not improve from 2.60055\n","Epoch 181/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4069 - val_loss: 2.6283\n","\n","Epoch 00181: val_loss did not improve from 2.60055\n","Epoch 182/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4047 - val_loss: 2.6456\n","\n","Epoch 00182: val_loss did not improve from 2.60055\n","Epoch 183/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4049 - val_loss: 2.6289\n","\n","Epoch 00183: val_loss did not improve from 2.60055\n","Epoch 184/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4000 - val_loss: 2.6229\n","\n","Epoch 00184: val_loss did not improve from 2.60055\n","Epoch 185/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4000 - val_loss: 2.6418\n","\n","Epoch 00185: val_loss did not improve from 2.60055\n","Epoch 186/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4085 - val_loss: 2.6350\n","\n","Epoch 00186: val_loss did not improve from 2.60055\n","Epoch 187/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4021 - val_loss: 2.6268\n","\n","Epoch 00187: val_loss did not improve from 2.60055\n","Epoch 188/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4056 - val_loss: 2.6342\n","\n","Epoch 00188: val_loss did not improve from 2.60055\n","Epoch 189/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4066 - val_loss: 2.6307\n","\n","Epoch 00189: val_loss did not improve from 2.60055\n","Epoch 190/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4055 - val_loss: 2.6493\n","\n","Epoch 00190: val_loss did not improve from 2.60055\n","Epoch 191/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4041 - val_loss: 2.6311\n","\n","Epoch 00191: val_loss did not improve from 2.60055\n","Epoch 192/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.4067 - val_loss: 2.6257\n","\n","Epoch 00192: val_loss did not improve from 2.60055\n","Epoch 193/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4069 - val_loss: 2.6267\n","\n","Epoch 00193: val_loss did not improve from 2.60055\n","Epoch 194/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.4035 - val_loss: 2.6357\n","\n","Epoch 00194: val_loss did not improve from 2.60055\n","Epoch 195/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4027 - val_loss: 2.6222\n","\n","Epoch 00195: val_loss did not improve from 2.60055\n","Epoch 196/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4046 - val_loss: 2.6349\n","\n","Epoch 00196: val_loss did not improve from 2.60055\n","Epoch 197/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4073 - val_loss: 2.6416\n","\n","Epoch 00197: val_loss did not improve from 2.60055\n","Epoch 198/200\n","179364/179364 [==============================] - 11s 63us/step - loss: 2.3986 - val_loss: 2.6323\n","\n","Epoch 00198: val_loss did not improve from 2.60055\n","Epoch 199/200\n","179364/179364 [==============================] - 11s 61us/step - loss: 2.3988 - val_loss: 2.6275\n","\n","Epoch 00199: val_loss did not improve from 2.60055\n","Epoch 200/200\n","179364/179364 [==============================] - 11s 62us/step - loss: 2.4025 - val_loss: 2.6371\n","\n","Epoch 00200: val_loss did not improve from 2.60055\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"LtzN-os2WSA2","colab_type":"text"},"source":["Loading the best model"]},{"cell_type":"code","metadata":{"id":"BwzKUEKYWS2s","colab_type":"code","colab":{}},"source":["model = load_model('drive/My Drive/ECE496 CNN/Saved Models/best_model.h5')"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"kANXYsbbh3ty","colab_type":"text"},"source":["Instantiate and train the model with a batch size of 128:"]},{"cell_type":"code","metadata":{"id":"XyX5drIvG0tC","colab_type":"code","colab":{}},"source":["#model = simple_wavenet()\n","#model.fit(X,np.array(y), epochs=40, batch_size=128,callbacks=[mc])"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"QZKQAUM3YzYg","colab_type":"text"},"source":["The steps mentioned under the inference phase for the predictions:"]},{"cell_type":"code","metadata":{"id":"elnilvIHYxQO","colab_type":"code","outputId":"452a34b5-6a16-464c-fadf-b7453572a285","executionInfo":{"status":"ok","timestamp":1587370416537,"user_tz":240,"elapsed":850,"user":{"displayName":"Sadman Sakif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhFzTKph8jrvXjeSmjPv-AWjlCXPOb8wSzBHYRB7w=s64","userId":"15126335579010581692"}},"colab":{"base_uri":"https://localhost:8080/","height":32}},"source":["ind = np.random.randint(0,len(x_val)-1)\n","\n","random_music = x_val[ind]\n","\n","#random_music = x_val[1350]\n","\n","predictions=[]\n","for i in range(10):\n","\n","    random_music = random_music.reshape(1,no_of_timesteps)\n","\n","    prob  = model.predict(random_music)[0]\n","    y_pred= np.argmax(prob,axis=0)\n","    predictions.append(y_pred)\n","\n","    random_music = np.insert(random_music[0],len(random_music[0]),y_pred)\n","    random_music = random_music[1:]\n","    \n","print(predictions)"],"execution_count":67,"outputs":[{"output_type":"stream","text":["[4, 6, 6, 6, 6, 51, 171, 51, 51, 6]\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"0lXOReGqZjfm","colab_type":"text"},"source":["Convert the integers back into the notes:"]},{"cell_type":"code","metadata":{"id":"Na6bhblfZnAo","colab_type":"code","colab":{}},"source":["x_int_to_note = dict((number, note_) for number, note_ in enumerate(unique_x)) \n","predicted_notes = [x_int_to_note[i] for i in predictions]"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"HtwfHYvBZqCt","colab_type":"text"},"source":["Convert back the predictions into a MIDI file:"]},{"cell_type":"code","metadata":{"id":"GEwb4D7Ya0wa","colab_type":"code","colab":{}},"source":["def convert_to_midi(prediction_output):\n","\n","  offset = 0\n","  output_notes = []\n","\n","  # create note and chord objects based on the values generated by the model\n","  for pattern in prediction_output:\n","    # pattern is a chord\n","    if ('.' in pattern) or pattern.isdigit():\n","      notes_in_chord = pattern.split('.')\n","      notes = []\n","\n","      for current_note in notes_in_chord:\n","        cn=int(current_note)\n","        new_note = note.Note(cn)\n","        new_note.storedInstrument = instrument.Piano()\n","        notes.append(new_note)\n","      \n","      new_chord = chord.Chord(notes)\n","      new_chord.offset = offset\n","      output_notes.append(new_chord)\n","\n","    # pattern is a note\n","    else:\n","      new_note = note.Note(pattern)\n","      new_note.offset = offset\n","      new_note.storedInstrument = instrument.Piano()\n","      output_notes.append(new_note)\n","\n","    # increase offset each iteration so that notes do not stack\n","    offset += 1\n","\n","  midi_stream = stream.Stream(output_notes)\n","  midi_stream.write('midi', fp='drive/My Drive/ECE496 CNN/Output/music_new.midi')\n"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"MRpu9a3YceYH","colab_type":"text"},"source":["Converting the predictions into a musical file:"]},{"cell_type":"code","metadata":{"id":"dlO1zYLicflA","colab_type":"code","colab":{}},"source":["convert_to_midi(predicted_notes)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"32AaDYYdu54c","colab_type":"text"},"source":["USER INPUT:"]},{"cell_type":"code","metadata":{"id":"gvPpWqFQu8Qn","colab_type":"code","outputId":"3489931f-b1df-448e-df32-70d656484ce0","executionInfo":{"status":"ok","timestamp":1587377164361,"user_tz":240,"elapsed":1681,"user":{"displayName":"Sadman Sakif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhFzTKph8jrvXjeSmjPv-AWjlCXPOb8wSzBHYRB7w=s64","userId":"15126335579010581692"}},"colab":{"base_uri":"https://localhost:8080/","height":387}},"source":["drive_path2    = 'drive/My Drive/ECE496 CNN/'\n","data_in_path2  = 'Input/'\n","output_path2 ='Output_new/'\n","\n","midi_dir2 = drive_path + data_in_path\n","out_dir2 = drive_path + output_path\n","\n","\n","\n","all_notes2=[]\n","\n","all_notes2.append(np.array(read_midi(\"drive/My Drive/ECE496 CNN/Input/input.midi\")))\n","\n","\n","#notes and chords of all the midi files\n","notes2 = [element for notes2 in all_notes2 for element in notes2]\n","\n","\n","unique_notes2 = list(set(notes2))\n","freq2 = dict(Counter(notes2))\n","\n","#consider only the frequencies\n","no2=[count for _,count in freq2.items()]\n","\n","#set the figure size\n","plt.figure(figsize=(5,5))\n","\n","#plot\n","plt.hist(no2)\n","\n"],"execution_count":177,"outputs":[{"output_type":"stream","text":["\n"],"name":"stdout"},{"output_type":"execute_result","data":{"text/plain":["(array([67., 28., 18.,  9.,  4.,  6.,  4.,  3.,  1.,  3.]),\n"," array([ 1. ,  8.6, 16.2, 23.8, 31.4, 39. , 46.6, 54.2, 61.8, 69.4, 77. ]),\n"," <a list of 10 Patch objects>)"]},"metadata":{"tags":[]},"execution_count":177},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 360x360 with 1 Axes>"]},"metadata":{"tags":[],"image/png":{"width":316,"height":305},"needs_background":"light"}}]},{"cell_type":"code","metadata":{"id":"L2ywbJ6wCiab","colab_type":"code","colab":{}},"source":["frequent_notes2 = [notes2 for notes2, count in freq2.items() if count>=50]\n","#print(len(frequent_notes2))"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"NlBrTnHvDFu2","colab_type":"code","colab":{}},"source":["new_music2 = []\n","for elm in all_notes2:\n","  temp = []\n","  for a in elm:\n","    if a in frequent_notes2:\n","      temp.append(a)\n","  new_music2.append(temp)\n","\n","new_music2 = np.array(new_music2)\n","\n","\n","#length of a input sequence\n","no_of_timesteps = 32\n","\n","x2 = []\n","#y = []\n","\n","for elm in new_music2:\n","  for i in range(0, len(elm) - no_of_timesteps, 1):\n","\n","    #preparing input and output sequences\n","    inp = elm[i:i + no_of_timesteps]\n","    #out = elm[i + no_of_timesteps]\n","\n","    x2.append(inp)\n","    #y.append(out)\n","\n","x2 = np.array(x2)\n","#y = np.array(y)\n","\n","\n","unique_x2 = list(set(x2.ravel()))\n","x_note_to_int2 = dict((note_, number) for number, note_ in enumerate(unique_x2))\n","\n","x_seq2 = []\n","for i in x2:\n","  temp = []\n","  for j in i:\n","    temp.append(x_note_to_int2[j])\n","  x_seq2.append(temp)\n","\n","x_seq2 = np.array(x_seq2)"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"YT4KJo3VJeBb","colab_type":"code","colab":{}},"source":["#len(x_seq2)"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"y932UKO9Ewuy","colab_type":"code","colab":{}},"source":["random_music2 = x_seq2[300]\n","\n","predictions2=[]\n","for i in range(20):\n","\n","    random_music2 = random_music2.reshape(1,no_of_timesteps)\n","\n","    prob  = model.predict(random_music2)[0]\n","    y_pred= np.argmax(prob,axis=0)\n","    predictions2.append(y_pred)\n","\n","    random_music2 = np.insert(random_music2[0],len(random_music2[0]),y_pred)\n","    random_music2 = random_music2[1:]\n","    \n","#print(predictions2)"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"6wvhMTWwFfT0","colab_type":"code","colab":{}},"source":["#x_int_to_note2 = dict((number, note_) for number, note_ in enumerate(unique_x2)) "],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"FlmvR56aH51T","colab_type":"code","colab":{}},"source":["predicted_notes2 = [x_int_to_note[i] for i in predictions2]"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"a_dx2hXeGOWU","colab_type":"code","colab":{}},"source":["def convert_to_midi2(prediction_output):\n","\n","  offset = 0\n","  output_notes = []\n","\n","  # create note and chord objects based on the values generated by the model\n","  for pattern in prediction_output:\n","    # pattern is a chord\n","    if ('.' in pattern) or pattern.isdigit():\n","      notes_in_chord = pattern.split('.')\n","      notes = []\n","\n","      for current_note in notes_in_chord:\n","        cn=int(current_note)\n","        new_note = note.Note(cn)\n","        new_note.storedInstrument = instrument.Piano()\n","        notes.append(new_note)\n","      \n","      new_chord = chord.Chord(notes)\n","      new_chord.offset = offset\n","      output_notes.append(new_chord)\n","\n","    # pattern is a note\n","    else:\n","      new_note = note.Note(pattern)\n","      new_note.offset = offset\n","      new_note.storedInstrument = instrument.Piano()\n","      output_notes.append(new_note)\n","\n","    # increase offset each iteration so that notes do not stack\n","    offset += 1\n","\n","  midi_stream = stream.Stream(output_notes)\n","  midi_stream.write('midi', fp='drive/My Drive/ECE496 CNN/Output_new/final.midi')"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"RhGH59ppGbWS","colab_type":"code","colab":{}},"source":["convert_to_midi2(predicted_notes2)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"8MKN0kDEcmAG","colab_type":"text"},"source":["#\n","#\n","#\n","#\n","#\n","#\n","#\n","#\n","#\n","# # # # # # # # #\n","# # # # # # # # #\n","# # # # # # # # #\n","# # # # # # # # #\n","# # # # # # # # #\n","# # # # # # # # #\n","# # # # # # # # #\n","# # # # # # # # #\n","# # # # # # # # #\n","# # # # # # # # #\n"]},{"cell_type":"markdown","metadata":{"id":"EDT9DjTsh-Td","colab_type":"text"},"source":["This is an implementation of the inference phase. It predicts the most likely element for a certain number of iterations:"]},{"cell_type":"code","metadata":{"id":"RuRkEWDVWaej","colab_type":"code","colab":{}},"source":["def generate_music(model, pitch, no_of_timesteps, pattern):\n","    \n","    int_to_note = dict((number, note) for number, note in enumerate(pitch))\n","    prediction_output = []\n","    \n","    # generate 50 elements\n","    for note_index in range(50):\n","        #reshaping array to feed into model\n","        input_ = np.reshape(pattern, (1, len(pattern), 1))\n","        #predict the probability and choose the maximum value\n","        proba = model.predict(input_, verbose=0)\n","        index = np.argmax(proba)\n","        #convert integer back to the element\n","        pred = int_to_note[index]\n","        prediction_output.append(pred)\n","        pattern = list(pattern)\n","        pattern.append(index/float(n_vocab))\n","        #leave the first value at index 0\n","        pattern = pattern[1:len(pattern)]\n","\n","    return prediction_output"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"XdesNjQniCh_","colab_type":"text"},"source":["Here is a function to convert the composed music into a MIDI file:"]},{"cell_type":"code","metadata":{"id":"PxgQaz6AiE4p","colab_type":"code","colab":{}},"source":["#def convert_to_midi(prediction_output):\n","#    offset = 0\n","#    output_notes = []\n","\n","#    # create note and chord objects based on the values generated by the model\n","#    for pattern in prediction_output:\n","        # pattern is a chord\n","#        if ('.' in pattern) or pattern.isdigit():\n","#            notes_in_chord = pattern.split('.')\n","#            notes = []\n","\n","#            for current_note in notes_in_chord:\n","#                new_note = note.Note(int(current_note))\n","#                new_note.storedInstrument = instrument.Piano()\n","#                notes.append(new_note)\n","#            new_chord = chord.Chord(notes)\n","#            new_chord.offset = offset\n","#            output_notes.append(new_chord)\n","        # pattern is a note\n","#        else:\n","#            new_note = note.Note(pattern)\n","#            new_note.offset = offset\n","#            new_note.storedInstrument = instrument.Piano()\n","#            output_notes.append(new_note)\n","\n","        # Specify duration between 2 notes\n","#        offset = offset + 0.5\n","       # offset += random.uniform(0.5,0.9)\n","\n","#    midi_stream = stream.Stream(output_notes)\n","#    midi_stream.write('midi', fp='drive/My Drive/ECE496 CNN/Output/music.midi')\n","#    return None"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"3itPJg-oDl93","colab_type":"code","outputId":"95be990c-f02e-4ec3-f85f-2ec7f25be241","executionInfo":{"status":"error","timestamp":1587370635978,"user_tz":240,"elapsed":2918,"user":{"displayName":"Sadman Sakif","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhFzTKph8jrvXjeSmjPv-AWjlCXPOb8wSzBHYRB7w=s64","userId":"15126335579010581692"}},"colab":{"base_uri":"https://localhost:8080/","height":323}},"source":["#Select random chunk for the first iteration\n","start = np.random.randint(0, len(x)-1)\n","pattern = x[start]\n","#pattern = X[50000]\n","#load the best model\n","#model=load_model('drive/My Drive/ECE496 CNN/Saved Models/model040.h5')\n","model = load_model('drive/My Drive/ECE496 CNN/Saved Models/best_model.h5')\n","#generate and save music\n","music = generate_music(model,pitch,no_of_timesteps,pattern)\n","convert_to_midi(music)"],"execution_count":74,"outputs":[{"output_type":"error","ename":"TypeError","evalue":"ignored","traceback":["\u001b[0;31m---------------------------------------------------------------------------\u001b[0m","\u001b[0;31mTypeError\u001b[0m                                 Traceback (most recent call last)","\u001b[0;32m<ipython-input-74-8a9d0f7cd948>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0mmodel\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mload_model\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'drive/My Drive/ECE496 CNN/Saved Models/best_model.h5'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0;31m#generate and save music\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 8\u001b[0;31m \u001b[0mmusic\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mgenerate_music\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmodel\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mpitch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mno_of_timesteps\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mpattern\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      9\u001b[0m \u001b[0mconvert_to_midi\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmusic\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n","\u001b[0;32m<ipython-input-73-5ee5120712b0>\u001b[0m in \u001b[0;36mgenerate_music\u001b[0;34m(model, pitch, no_of_timesteps, pattern)\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0mgenerate_music\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmodel\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpitch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mno_of_timesteps\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpattern\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m     \u001b[0mint_to_note\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnumber\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnote\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mnumber\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnote\u001b[0m \u001b[0;32min\u001b[0m \u001b[0menumerate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpitch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m     \u001b[0mprediction_output\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n","\u001b[0;31mTypeError\u001b[0m: 'module' object is not iterable"]}]},{"cell_type":"code","metadata":{"id":"XQs4TvjLEBja","colab_type":"code","colab":{}},"source":["X[start].size"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"9saSv3sClfXF","colab_type":"code","colab":{}},"source":["len(all_notes[0])"],"execution_count":0,"outputs":[]}]}