32 lines
1.0 KiB
Python
32 lines
1.0 KiB
Python
|
import tensorflow as tf
|
||
|
|
import settings
|
||
|
|
from tensorflow.keras import layers
|
||
|
|
from keras.layers import Input, Dense, Conv2D, Flatten
|
||
|
|
from keras.models import Model, Sequential
|
||
|
|
import numpy as np
|
||
|
|
from sys import exit
|
||
|
|
import pickle
|
||
|
|
|
||
|
|
print('Reading samples from: {}'.format(settings.samples_path))
|
||
|
|
|
||
|
|
train_X = np.load(settings.samples_path)['arr_0']
|
||
|
|
|
||
|
|
n_samples = train_X.shape[0]
|
||
|
|
input_shape = settings.midi_resolution*128
|
||
|
|
train_X = train_X.reshape(n_samples, input_shape)
|
||
|
|
|
||
|
|
# encoder model
|
||
|
|
input_img = tf.keras.layers.Input(shape=(input_shape,))
|
||
|
|
encoded = tf.keras.layers.Dense(160, activation='relu')(input_img)
|
||
|
|
decoded = tf.keras.layers.Dense(input_shape, activation='sigmoid')(encoded)
|
||
|
|
autoencoder = tf.keras.models.Model(input_img, decoded)
|
||
|
|
|
||
|
|
autoencoder.compile(optimizer='adam',
|
||
|
|
loss='binary_crossentropy',
|
||
|
|
metrics=['accuracy'])
|
||
|
|
|
||
|
|
autoencoder.fit(train_X, train_X, epochs=settings.epochs, batch_size=32)
|
||
|
|
|
||
|
|
autoencoder.save_weights(settings.model_path)
|
||
|
|
print("Model save to {}".format(settings.model_path))
|