final model

backend/backend/cms/apps.py

@@ -3,4 +3,4 @@ from django.apps import AppConfig
 
 class CmsConfig(AppConfig):
     default_auto_field = 'django.db.models.BigAutoField'
-    name = 'cms'
+    name = 'backend.cms'

backend/backend/cms/migrations/0001_initial.py

+# Generated by Django 4.0.4 on 2022-04-25 19:51
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    initial = True
+
+    dependencies = [
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name='MlModel',
+            fields=[
+                ('id', models.AutoField(primary_key=True, serialize=False)),
+                ('description', models.TextField(blank=True)),
+                ('timestamp', models.DateTimeField(auto_now=True)),
+                ('details', models.JSONField()),
+            ],
+        ),
+    ]

backend/backend/cms/model/predict.py

+import pickle
+from keras.models import load_model
+
+
+import numpy as np
+
+
+import IPython.display as ipd
+
+import random
+
+from sklearn.model_selection import train_test_split
+from sklearn.preprocessing import LabelEncoder
+
+
+def predict():
+    model=load_model(r'./best_model_final.hdf5')
+
+    f1 = open('all_label.txt', 'rb')
+    all_label = pickle.load(f1)
+    print('loaded labels')
+
+    f2 = open('all_waves_file.txt', 'rb')
+    all_wave = pickle.load(f2)
+    print('loaded waves')
+
+    le = LabelEncoder()
+    y = le.fit_transform(all_label)
+    classes = list(le.classes_)
+
+    train_data_file = open("train_data_file.txt", 'rb')
+    [x_tr, x_val, y_tr, y_val] = np.load(train_data_file, allow_pickle=True)
+    train_data_file.close()
+
+    def predictSamples(audio):
+        prob=model.predict(audio.reshape(1,8000,1))
+        index=np.argmax(prob[0])
+        return classes[index]
+
+        
+    index=random.randint(0,len(x_val)-1)
+    samples=x_val[index].ravel()
+    print("Audio:",classes[np.argmax(y_val[index])])
+    ipd.Audio(samples, rate=8000)
+
+    print("Text:",predictSamples(samples))
\ No newline at end of file

backend/backend/cms/model/train.py

+import pickle
+from matplotlib import pyplot
 import os
 import librosa
 import IPython.display as ipd
@@ -6,84 +8,192 @@ import numpy as np
 from scipy.io import wavfile
 import warnings
 
-
 from sklearn.preprocessing import LabelEncoder
 
 from keras.utils import np_utils
 
+from sklearn.model_selection import train_test_split
+
+from keras.layers import Dense, Dropout, Flatten, Conv1D, Input, MaxPooling1D
+from keras.models import Model
+from keras.callbacks import EarlyStopping, ModelCheckpoint
+from keras import backend as K
+K.clear_session()
+
+
 warnings.filterwarnings("ignore")
 
-os.listdir('../input/')
+# os.listdir('../../../data/')
+
+classes = ['down', 'go', 'left', 'no', 'off',
+           'on', 'right', 'stop', 'up', 'yes']
+
+
+def train():
+
+    print('1')
+    train_audio_path = r'./backend/data/train/train/audio/'
+    samples, sample_rate = librosa.load(
+        train_audio_path+'yes/0a7c2a8d_nohash_0.wav', sr=16000)
+    # fig = plt.figure(figsize=(14, 8))
+    # ax1 = fig.add_subplot(211)
+    # ax1.set_title('Raw wave of ' + r'../input/train/audio/yes/0a7c2a8d_nohash_0.wav')
+    # ax1.set_xlabel('time')
+    # ax1.set_ylabel('Amplitude')
+    # ax1.plot(np.linspace(0, sample_rate/len(samples), sample_rate), samples)
+
+    ipd.Audio(samples, rate=sample_rate)
+
+    print(sample_rate)
+
+    samples = librosa.resample(samples, sample_rate, 8000)
+    ipd.Audio(samples, rate=8000)
+
+    labels = os.listdir(train_audio_path)
+
+    # find count of each label and plot bar graph
+    no_of_recordings = []
+    for label in labels:
+        waves = [f for f in os.listdir(
+            train_audio_path + '/' + label) if f.endswith('.wav')]
+        no_of_recordings.append(len(waves))
+
+    # plot
+    # plt.figure(figsize=(30,5))
+    index = np.arange(len(labels))
+    # plt.bar(index, no_of_recordings)
+    # plt.xlabel('Commands', fontsize=12)
+    # plt.ylabel('No of recordings', fontsize=12)
+    # plt.xticks(index, labels, fontsize=15, rotation=60)
+    # plt.title('No. of recordings for each command')
+    # plt.show()
+
+    print('2')
+
+    labels = ["yes", "no", "up", "down", "left",
+              "right", "on", "off", "stop", "go"]
+
+    # labels_file = open('./labels_file.bin', 'wb+')
+    # pickle.dump(obj=labels, file=labels_file)
+    # labels_file.close()
+
+    # # file = open('./labels_file.bin', 'rb')
+    # # dict = pickle.load(file)
+    # # print('loaded')
+    # # print(dict)
+    # # print('fdnasf')
+
+    duration_of_recordings = []
+    for label in labels:
+        print('2.1', label)
+        waves = [f for f in os.listdir(
+            train_audio_path + '/' + label) if f.endswith('.wav')]
+        for wav in waves:
+            sample_rate, samples = wavfile.read(
+                train_audio_path + '/' + label + '/' + wav)
+            duration_of_recordings.append(float(len(samples)/sample_rate))
+
+    plt.hist(np.array(duration_of_recordings))
+
+    train_audio_path = r'./backend/data/train/train/audio/'
+
+    # all_wave = []
+    # all_label = []
+    # for label in labels:
+    #     print(label)
+    #     waves = [f for f in os.listdir(
+    #         train_audio_path + '/' + label) if f.endswith('.wav')]
+    #     for wav in waves:
+    #         samples, sample_rate = librosa.load(
+    #             train_audio_path + '/' + label + '/' + wav, sr=16000)
+    #         samples = librosa.resample(samples, sample_rate, 8000)
+    #         if(len(samples) == 8000):
+    #             all_wave.append(samples)
+    #             all_label.append(label)
+
+    # print('3')
+
+    f1 = open('all_label.txt', 'rb')
+    all_label = pickle.load(f1)
+    print('loaded labels')
 
+    f2 = open('all_waves_file.txt', 'rb')
+    all_wave = pickle.load(f2)
+    print('loaded waves')
 
-train_audio_path = '../input/tensorflow-speech-recognition-challenge/train/audio/'
-samples, sample_rate = librosa.load(train_audio_path+'yes/0a7c2a8d_nohash_0.wav', sr = 16000)
-fig = plt.figure(figsize=(14, 8))
-ax1 = fig.add_subplot(211)
-ax1.set_title('Raw wave of ' + '../input/train/audio/yes/0a7c2a8d_nohash_0.wav')
-ax1.set_xlabel('time')
-ax1.set_ylabel('Amplitude')
-ax1.plot(np.linspace(0, sample_rate/len(samples), sample_rate), samples)
+    le = LabelEncoder()
+    y = le.fit_transform(all_label)
+    classes = list(le.classes_)
 
+    print(classes)
+    print(all_wave)
 
-ipd.Audio(samples, rate=sample_rate)
+    print('4')
+
+    y = np_utils.to_categorical(y, num_classes=len(labels))
+
+    all_wave = np.array(all_wave).reshape(-1, 8000, 1)
+
+    x_tr, x_val, y_tr, y_val = train_test_split(np.array(all_wave), np.array(
+        y), stratify=y, test_size=0.2, random_state=777, shuffle=True)
 
-print(sample_rate)
+    train_data_file = open('train_data_file.txt', 'wb+')
+    np.save(file=train_data_file, arr=np.array([x_tr, x_val, y_tr, y_val]))
+    train_data_file.close()
 
+    inputs = Input(shape=(8000, 1))
 
-samples = librosa.resample(samples, sample_rate, 8000)
-ipd.Audio(samples, rate=8000)
+    # First Conv1D layer
+    conv = Conv1D(8, 13, padding='valid', activation='relu', strides=1)(inputs)
+    conv = MaxPooling1D(3)(conv)
+    conv = Dropout(0.3)(conv)
 
+    # Second Conv1D layer
+    conv = Conv1D(16, 11, padding='valid', activation='relu', strides=1)(conv)
+    conv = MaxPooling1D(3)(conv)
+    conv = Dropout(0.3)(conv)
+
+    # Third Conv1D layer
+    conv = Conv1D(32, 9, padding='valid', activation='relu', strides=1)(conv)
+    conv = MaxPooling1D(3)(conv)
+    conv = Dropout(0.3)(conv)
 
-labels=os.listdir(train_audio_path)
+    # Fourth Conv1D layer
+    conv = Conv1D(64, 7, padding='valid', activation='relu', strides=1)(conv)
+    conv = MaxPooling1D(3)(conv)
+    conv = Dropout(0.3)(conv)
 
-#find count of each label and plot bar graph
-no_of_recordings=[]
-for label in labels:
-    waves = [f for f in os.listdir(train_audio_path + '/'+ label) if f.endswith('.wav')]
-    no_of_recordings.append(len(waves))
-    
-#plot
-plt.figure(figsize=(30,5))
-index = np.arange(len(labels))
-plt.bar(index, no_of_recordings)
-plt.xlabel('Commands', fontsize=12)
-plt.ylabel('No of recordings', fontsize=12)
-plt.xticks(index, labels, fontsize=15, rotation=60)
-plt.title('No. of recordings for each command')
-plt.show()
+    # Flatten layer
+    conv = Flatten()(conv)
 
-labels=["yes", "no", "up", "down", "left", "right", "on", "off", "stop", "go"]
+    # Dense Layer 1
+    conv = Dense(256, activation='relu')(conv)
+    conv = Dropout(0.3)(conv)
 
-duration_of_recordings=[]
-for label in labels:
-    waves = [f for f in os.listdir(train_audio_path + '/'+ label) if f.endswith('.wav')]
-    for wav in waves:
-        sample_rate, samples = wavfile.read(train_audio_path + '/' + label + '/' + wav)
-        duration_of_recordings.append(float(len(samples)/sample_rate))
-    
-plt.hist(np.array(duration_of_recordings))
+    # Dense Layer 2
+    conv = Dense(128, activation='relu')(conv)
+    conv = Dropout(0.3)(conv)
 
-train_audio_path = '../input/tensorflow-speech-recognition-challenge/train/audio/'
+    outputs = Dense(len(labels), activation='softmax')(conv)
 
-all_wave = []
-all_label = []
-for label in labels:
-    print(label)
-    waves = [f for f in os.listdir(train_audio_path + '/'+ label) if f.endswith('.wav')]
-    for wav in waves:
-        samples, sample_rate = librosa.load(train_audio_path + '/' + label + '/' + wav, sr = 16000)
-        samples = librosa.resample(samples, sample_rate, 8000)
-        if(len(samples)== 8000) : 
-            all_wave.append(samples)
-            all_label.append(label)
+    model = Model(inputs, outputs)
+    model.summary()
 
+    model.compile(loss='categorical_crossentropy',
+                  optimizer='adam', metrics=['accuracy'])
 
+    es = EarlyStopping(monitor='val_loss', mode='min',
+                       verbose=1, patience=10, min_delta=0.0001)
+    mc = ModelCheckpoint('best_model.hdf5', monitor='val_accuracy',
+                         verbose=1, save_best_only=True, mode='max')
 
+    history = model.fit(x_tr, y_tr, epochs=100, callbacks=[
+                        es, mc], batch_size=32, validation_data=(x_val, y_val))
 
-le = LabelEncoder()
-y=le.fit_transform(all_label)
-classes= list(le.classes_)
+    # pyplot.plot(history.history['loss'], label='train')
+    # pyplot.plot(history.history['val_loss'], label='test')
+    # pyplot.legend()
 
+    # pyplot.show()
 
-y=np_utils.to_categorical(y, num_classes=len(labels))
\ No newline at end of file
+    return history

backend/backend/cms/models.py

 from django.db import models
 
 # Create your models here.
+class MlModel(models.Model):
+    id = models.AutoField(primary_key=True)
+    description = models.TextField(blank=True)
+    timestamp = models.DateTimeField(blank=True, auto_now=True)
+    details = models.JSONField()

backend/backend/cms/serializers.py

 from django.contrib.auth.models import User, Group
+from .models import MlModel
 from rest_framework import serializers
 
 
@@ -11,4 +12,11 @@ class UserSerializer(serializers.HyperlinkedModelSerializer):
 class GroupSerializer(serializers.HyperlinkedModelSerializer):
     class Meta:
         model = Group
-        fields = ['url', 'name']
\ No newline at end of file
+        fields = ['url', 'name']
+
+class MlModelSerializer(serializers.ModelSerializer):
+    class Meta:
+        model = MlModel
+        fields = (
+            '__all__'
+        )
\ No newline at end of file

backend/backend/cms/views.py

@@ -2,10 +2,17 @@ from http.client import HTTPResponse
 from django.contrib.auth.models import User, Group
 from rest_framework import viewsets
 from rest_framework import permissions
+from backend.cms.serializers import MlModelSerializer
 from backend.cms.serializers import UserSerializer, GroupSerializer
 from rest_framework.decorators import action
 from rest_framework.response import Response
 
+from .models import MlModel
+
+from .model.train import train
+
+from .model.predict import predict
+
 
 class UserViewSet(viewsets.ModelViewSet):
     """
@@ -20,6 +27,9 @@ class UserViewSet(viewsets.ModelViewSet):
 
 
         print('Function ran')
+        results = train()
+        print(results)
+
         return Response({'success': True})
 
 
@@ -29,4 +39,29 @@ class GroupViewSet(viewsets.ModelViewSet):
     """
     queryset = Group.objects.all()
     serializer_class = GroupSerializer
-    permission_classes = [permissions.IsAuthenticated]
\ No newline at end of file
+    permission_classes = [permissions.IsAuthenticated]
+
+class MlModelViewSet(viewsets.ViewSet):
+
+    queryset = MlModel.objects.all()
+    serializer_class = MlModelSerializer
+    permission_classes = [permissions.IsAuthenticated]
+
+    @action(detail=False)
+    def train(*args, **kwargs):
+        print('Function ran')
+        results = train()
+        print(results)
+        return Response({'success': True})
+
+
+    @action(detail=False)
+    def predict(*args, **kwargs):
+        print('Function ran')
+        results = predict()
+        print(results)
+        return Response({'success': True})
+
+    
+
+    

backend/backend/settings.py

@@ -38,6 +38,7 @@ INSTALLED_APPS = [
     'django.contrib.messages',
     'django.contrib.staticfiles',
     'rest_framework',
+    'backend.cms'
 ]
 
 MIDDLEWARE = [

backend/backend/urls.py

@@ -5,6 +5,7 @@ from backend.cms import views
 router = routers.DefaultRouter()
 router.register(r'users', views.UserViewSet)
 router.register(r'groups', views.GroupViewSet)
+router.register(r'mlmodels', views.MlModelViewSet)
 
 # Wire up our API using automatic URL routing.
 # Additionally, we include login URLs for the browsable API.

requirements.txt

 absl-py==1.0.0