e18df168c65541107227e567de3a99cb21cfa426,theanolm/commands/decode.py,,decode,#Any#,91

Before Change

                     args.job)
        tokens = decoder.decode(lattice)

        best_token = tokens[0]
        words = vocabulary.id_to_word[best_token.history]
        if args.output == "ref":
            args.output_file.write("{} {}\n".format(utterance_id, " ".join(words)))
        elif args.output == "trn":
            args.output_file.write("{} ({})\n".format(" ".join(words), utterance_id))
        elif args.output == "n-best":
            ac_logprob = best_token.ac_logprob / log_scale
            lm_logprob = best_token.ac_logprob / log_scale
            args.output_file.write("{} {} {} {}\n".format(
                utterance_id, logprob, len(words), " ".join(words)))
        else:

After Change

                     args.job)
        tokens = decoder.decode(lattice)

        for index in range(min(args.n_best, len(tokens))):
            line = format_token(tokens[index],
                                utterance_id,
                                vocabulary,
                                log_scale,
                                args.output)
            args.output_file.write(line + "\n")

def format_token(token, utterance_id, vocabulary, log_scale, format):
    Formats an output line from a token and an utterance ID.
    

Instances