Upload Train.py

Train.py

+import os
+import numpy as np
+import librosa
+import tensorflow as tf
+from sklearn.model_selection import train_test_split
+
+# Define a function to extract features from the audio file
+def extract_features(audio_path):
+    # Load the audio file
+    y, sr = librosa.load(audio_path, sr=22050, duration=3)
+    # Extract features using Mel-frequency cepstral coefficients (MFCC)
+    mfcc = librosa.feature.mfcc(y=y, sr=sr, n_mfcc=20)
+    # Compute the mean of each feature dimension
+    mfcc_mean = np.mean(mfcc, axis=1)
+    # Compute the first and second derivatives of the MFCC features
+    mfcc_delta = librosa.feature.delta(mfcc)
+    mfcc_delta2 = librosa.feature.delta(mfcc, order=2)
+    # Compute the mean of each derivative feature dimension
+    mfcc_delta_mean = np.mean(mfcc_delta, axis=1)
+    mfcc_delta2_mean = np.mean(mfcc_delta2, axis=1)
+    # Concatenate the feature vectors
+    features = np.concatenate((mfcc_mean, mfcc_delta_mean, mfcc_delta2_mean))
+    return features
+
+# Define a function to load the audio files and their labels
+def load_data(data_dir):
+    audio_files = []
+    labels = []
+    # Loop through all subdirectories in the data directory
+    for root, dirs, files in os.walk(data_dir):
+        for file in files:
+            if file.endswith('.wav'):
+                # Extract the label from the filename
+                label = file.split('-')[2]
+                # Convert the label to an integer
+                label = int(label)
+                # Map the label to an emotion label
+                if label == 1:
+                    emotion = 'neutral'
+                elif label == 2:
+                    emotion = 'calm'
+                elif label == 3:
+                    emotion = 'happy'
+                elif label == 4:
+                    emotion = 'sad'
+                elif label == 5:
+                    emotion = 'angry'
+                elif label == 6:
+                    emotion = 'fearful'
+                elif label == 7:
+                    emotion = 'disgust'
+
+                # Add the audio file and its label to the lists
+                audio_files.append(os.path.join(root, file))
+                labels.append(emotion)
+    return audio_files, labels
+
+# Load the data
+data_dir = 'ravdess'
+audio_files, labels = load_data(data_dir)
+
+# Extract features from the audio files
+features = []
+for audio_file in audio_files:
+    feature = extract_features(audio_file)
+    features.append(feature)
+features = np.array(features)
+
+# Map the emotion labels to integer values
+label_map = {'neutral': 0, 'calm': 1, 'happy': 2, 'sad': 3, 'angry': 4, 'fearful': 5, 'disgust': 6}
+labels = np.array([label_map[label] for label in labels])
+
+# Split the data into training and testing sets
+X_train, X_test, y_train, y_test = train_test_split(features, labels, test_size=0.2, random_state=42)
+
+model = tf.keras.Sequential([
+tf.keras.layers.Input(shape=(60,)),
+tf.keras.layers.Dense(64, activation='relu'),
+tf.keras.layers.Dense(32, activation='relu'),
+tf.keras.layers.Dense(7, activation='softmax')
+])
+
+model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])
+
+model.fit(X_train, y_train, validation_data=(X_test, y_test), epochs=50, batch_size=32)
+
+model_path = 'model.h5'
+model.save(model_path)
+print('Model saved at:', model_path)
+
+test_loss, test_acc = model.evaluate(X_test, y_test, verbose=2)
+print('Test accuracy:', test_acc)
\ No newline at end of file