import numpy as np
import tensorflow as tf
from keras.layers import Input, Dense, Conv2D
from keras.models import Model
import settings

#model
input_shape = settings.midi_resolution*128
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='adadelta',
              loss='categorical_crossentropy',
              metrics=['accuracy'])

# load weights into new model
autoencoder.load_weights(settings.model_path)
print("Loaded model from {}".format(settings.model_path))

# generate_seed = np.random.rand(12288).reshape(1,12288)
generate_seed = np.load(settings.samples_path)['arr_0'][15].reshape(1,12288)

generated_sample = autoencoder.predict(generate_seed)
np.savez_compressed(settings.generated_sample_path, generated_sample)