def plot_images_encoded_in_latent_space(latent_representations, sample_labels):
plt.figure(figsize=(10, 10))
plt.scatter(latent_representations[:, 0],
latent_representations[:, 1],
cmap="rainbow",
c=sample_labels,
alpha=0.5,
s=2)
plt.colorbar()
plt.show()
if __name__ == "__main__":
autoencoder = VAE.load("model")
x_train, y_train, x_test, y_test = load_mnist()
num_sample_images_to_show = 8
sample_images, _ = select_images(x_test, y_test, num_sample_images_to_show)
reconstructed_images, _ = autoencoder.reconstruct(sample_images)
plot_reconstructed_images(sample_images, reconstructed_images)
num_images = 6000
sample_images, sample_labels = select_images(x_test, y_test, num_images)
_, latent_representations = autoencoder.reconstruct(sample_images)
plot_images_encoded_in_latent_space(latent_representations, sample_labels)
min_max_values[file_path] for file_path in file_paths
]
print(file_paths)
print(sampled_min_max_values)
return sampled_spectrogrmas, sampled_min_max_values
def save_signals(signals, save_dir, sample_rate=22050):
for i, signal in enumerate(signals):
save_path = os.path.join(save_dir, str(i) + ".wav")
sf.write(save_path, signal, sample_rate)
if __name__ == "__main__":
# initialise sound generator
vae = VAE.load("model")
sound_generator = SoundGenerator(vae, HOP_LENGTH)
# load spectrograms + min max values
with open(MIN_MAX_VALUES_PATH, "rb") as f:
min_max_values = pickle.load(f)
specs, file_paths = load_fsdd(SPECTROGRAMS_PATH)
# sample spectrograms + min max values
sampled_specs, sampled_min_max_values = select_spectrograms(
specs, file_paths, min_max_values, 5)
# generate audio for sampled spectrograms
signals, _ = sound_generator.generate(sampled_specs,
sampled_min_max_values)
