c343409098b4f4b8396119d9f26e040e479a0e2b,labs/08/speech_recognition_eval.py,,,#,18

Before Change

args = parser.parse_args()

with open(args.system, "r", encoding="utf-8") as system_file:
    system = [line.rstrip("\n") for line in system_file]

with open(args.gold, "r", encoding="utf-8") as gold_file:
    gold = [line.rstrip("\n") for line in gold_file]

After Change

            )
    return a[-1][-1]

if __name__ == "__main__":
    // Parse arguments
    parser = argparse.ArgumentParser()
    parser.add_argument("predictions", type=str, help="Path to predicted output.")
    parser.add_argument("dataset", type=str, help="Which dataset to evaluate ("dev", "test").")
    args = parser.parse_args([] if "__file__" not in globals() else None)

    gold = getattr(TimitMFCC(), args.dataset).data["letters"]

    with open(args.predictions, "r", encoding="utf-8") as predictions_file:
        predictions = [line.rstrip("\n") for line in predictions_file]

    if len(predictions) < len(gold):
        raise RuntimeError("The predictions are shorter than gold data: {} vs {}.".format(len(predictions), len(gold)))

    score = 0
    for i in range(len(gold)):
        gold_sentence = [TimitMFCC.LETTERS[letter] for letter in gold[i]]
        predicted_sentence = predictions[i].split(" ")
        score += edit_distance(gold_sentence, predicted_sentence) / len(gold_sentence)

    print("Average normalized edit distance: {:.2f}%".format(100 * score / len(gold)))

Instances