Project initilization

config.py

@@ -0,0 +1,5 @@
+# Configuration for the music generation model
+BATCH_SIZE = 64
+EPOCHS = 100
+LEARNING_RATE = 0.001
+MAX_SEQUENCE_LENGTH = 50

dataset/midi_utils.py

@@ -0,0 +1,25 @@
+import music21
+from midiutil import MIDIFile
+
+def midi_to_notes(midi_file):
+    notes = []
+    midi = music21.converter.parse(midi_file)
+    for element in midi.flat.notes:
+        if isinstance(element, music21.note.Note):
+            notes.append(str(element.pitch))  # could convert to MIDI number here if needed
+    return notes
+
+def create_midi_file(notes, filename):
+    midi = MIDIFile(1)
+    track = 0
+    time = 0
+
+    midi.addTrackName(track, time, "Generated Music")
+    midi.addTempo(track, time, 60)
+
+    for note in notes:
+        midi.addNote(track, 0, int(note), time, 1, 100)  # Ensure notes are in int format
+        time += 1
+
+    with open(filename, "wb") as f:
+        midi.writeFile(f)

generate_music.py

@@ -0,0 +1,18 @@
+import numpy as np
+from keras.models import load_model
+from midi_utils import create_midi_file
+
+def generate_music(model, start_seed, num_notes=100):
+    generated = start_seed
+    for _ in range(num_notes):
+        input_sequence = np.array(generated[-50:]).reshape(1, 50)
+        predicted = model.predict(input_sequence, verbose=0)
+        next_note = np.argmax(predicted)
+        generated.append(next_note)
+    return generated
+
+if __name__ == "__main__":
+    model = load_model('music_model.h5')
+    start_seed = [60, 62, 64]  # Example seed (MIDI note values)
+    generated_notes = generate_music(model, start_seed)
+    create_midi_file(generated_notes, 'generated_music.mid')

midi_utils.py

@@ -0,0 +1,25 @@
+import music21
+from midiutil import MIDIFile
+
+def midi_to_notes(midi_file):
+    notes = []
+    midi = music21.converter.parse(midi_file)
+    for element in midi.flat.notes:
+        if isinstance(element, music21.note.Note):
+            notes.append(str(element.pitch))  # could convert to MIDI number here if needed
+    return notes
+
+def create_midi_file(notes, filename):
+    midi = MIDIFile(1)
+    track = 0
+    time = 0
+
+    midi.addTrackName(track, time, "Generated Music")
+    midi.addTempo(track, time, 60)
+
+    for note in notes:
+        midi.addNote(track, 0, int(note), time, 1, 100)  # Ensure notes are in int format
+        time += 1
+
+    with open(filename, "wb") as f:
+        midi.writeFile(f)

models.py

@@ -0,0 +1,12 @@
+from keras.models import Sequential
+from keras.layers import LSTM, Dense, Embedding
+
+def create_model(vocab_size, max_sequence_length):
+    model = Sequential()
+    model.add(Embedding(input_dim=vocab_size, output_dim=256, input_length=max_sequence_length))
+    model.add(LSTM(256, return_sequences=True))
+    model.add(LSTM(256))
+    model.add(Dense(vocab_size, activation='softmax'))
+
+    model.compile(loss='categorical_crossentropy', optimizer='adam')
+    return model

requirements.txt

@@ -0,0 +1,5 @@
+numpy
+tensorflow>=2.0.0
+keras
+music21
+midiutil

train_model.py

@@ -0,0 +1,30 @@
+import numpy as np
+from keras.models import Sequential
+from keras.layers import LSTM, Dense, Embedding
+from keras.utils import to_categorical
+from midi_utils import midi_to_notes
+
+def prepare_data(notes):
+    # Convert notes to integers or one-hot encoding
+    # This function must be implemented.
+    # Placeholder: Assume we return X, y from preprocessing
+    pass 
+
+def create_model(vocab_size, max_sequence_length):
+    model = Sequential()
+    model.add(Embedding(input_dim=vocab_size, output_dim=256, input_length=max_sequence_length))
+    model.add(LSTM(256, return_sequences=True))
+    model.add(LSTM(256))
+    model.add(Dense(vocab_size, activation='softmax'))
+    model.compile(loss='categorical_crossentropy', optimizer='adam')
+    return model
+
+def train_model():
+    notes = midi_to_notes('dataset/your_midi_file1.mid')  # Load your dataset
+    X, y = prepare_data(notes)  # Process your data
+    model = create_model(vocab_size=len(set(notes)), max_sequence_length=50)  # Adjust variables as needed
+    model.fit(X, to_categorical(y), epochs=100, batch_size=64)
+    model.save('music_model.h5')
+
+if __name__ == "__main__":
+    train_model()