def main(argv):
if len(argv) > 1:
raise app.UsageError('Too many command-line arguments.')
dataset = preprocessor.get_dataset_from_tfds(FLAGS.dataset, FLAGS.split)
preprocessor.write_dataset(dataset, FLAGS.save_path)
token_vocab = preprocessor.get_token_vocab(FLAGS.save_path)
preprocessor.write_token_vocab(token_vocab, FLAGS.save_path)
"hertz_high" : args.hertz_high,
"normalize_mel" : args.normalize_mel,
"max_duration" : args.max_duration
}
# Create dataset
train_dataset_list = ["train-clean-100", "train-clean-360"]
train_dataset = preprocess_librispeech(args.data_dir, train_dataset_list, hp)
dev_dataset_list = ["dev-clean"]
dev_dataset = preprocess_librispeech(args.data_dir, dev_dataset_list, hp)
# Serialize dataset
train_dataset = train_dataset.map(
preprocess.serialize_example,
num_parallel_calls=tf.data.experimental.AUTOTUNE)
dev_dataset = dev_dataset.map(
preprocess.serialize_example,
num_parallel_calls=tf.data.experimental.AUTOTUNE)
# Store dataset
train_dataset_path = os.path.join(args.out_dir, "train.tfrecord")
os.makedirs(args.out_dir, exist_ok=True)
preprocess.write_dataset(train_dataset, train_dataset_path)
print("preprocessing for train dataset is done")
dev_dataset_path = os.path.join(args.out_dir, "dev.tfrecord")
os.makedirs(args.out_dir, exist_ok=True)
preprocess.write_dataset(dev_dataset, dev_dataset_path)
print("preprocessing for dev dataset is done")