if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("-l", "--linguistic_model", type=str, required=True)
parser.add_argument("-a", "--acoustic_model", type=str, required=True)
args = parser.parse_args()
assert isfile(
args.acoustic_model), "acoustic_model weights file does not exist"
assert isfile(args.acoustic_model.replace(
".torch", ".json")), "acoustic_model config file does not exist"
assert isfile(
args.linguistic_model), "linguistic_model weights file does not exist"
assert isfile(args.linguistic_model.replace(
".torch", ".json")), "linguistic_model config file does not exist"
test_features_acoustic, test_labels_acoustic, val_features_acoustic, val_labels_acoustic, _, _ = load_spectrogram_dataset(
)
test_iterator_acoustic = BatchIterator(test_features_acoustic,
test_labels_acoustic, 100)
test_features_linguistic, test_labels_linguistic, val_features_linguistic, val_labels_linguistic, _, _ = load_linguistic_dataset(
)
test_iterator_linguistic = BatchIterator(test_features_linguistic,
test_labels_linguistic, 100)
val_iterator_acoustic = BatchIterator(val_features_acoustic,
val_labels_acoustic, 100)
val_iterator_linguistic = BatchIterator(val_features_linguistic,
val_labels_linguistic, 100)
assert np.array_equal(
test_labels_acoustic, test_labels_linguistic
), "Labels for acoustic and linguistic datasets are not the same!"
"""Choosing hardware"""
parser.add_argument("-m", "--model_type", type=str, default="linguistic")
args = parser.parse_args()
if args.model_type == "linguistic":
cfg = LinguisticConfig()
test_features, test_labels, val_features, val_labels, train_features, train_labels = load_linguistic_dataset(
)
model = RNN(cfg)
elif args.model_type == "acoustic-lld":
cfg = AcousticLLDConfig()
test_features, test_labels, val_features, val_labels, train_features, train_labels = load_acoustic_features_dataset(
)
model = RNN(cfg)
elif args.model_type == "acoustic-spectrogram":
cfg = AcousticSpectrogramConfig()
test_features, test_labels, val_features, val_labels, train_features, train_labels = load_spectrogram_dataset(
)
model = CNN(cfg)
else:
raise Exception(
"model_type parameter has to be one of [linguistic|acoustic-lld|acoustic-spectrogram]"
)
print(
"Subsets sizes: test_features:{}, test_labels:{}, val_features:{}, val_labels:{}, train_features:{}, train_labels:{}"
.format(test_features.shape[0], test_labels.shape[0],
val_features.shape[0], val_labels.shape[0],
train_features.shape[0], train_labels.shape[0]))
"""Running training"""
run_training(model, cfg, test_features, test_labels, train_features,
train_labels, val_features, val_labels)