志浩
2023-02-15 68b87c91405e11dca5dc64ef1e1b2fdc3e2389f4
egs/mars/sd/scripts/dump_rttm_to_labels.py
@@ -31,9 +31,25 @@
        for scp_path in wav_scp_list:
            meeting_scp.update(load_scp_as_dict(scp_path))
        assert len(meeting_scp) == len(meeting2rttm), \
            "Number of wav and rttm mismatch {} != {}".format(len(meeting_scp), len(meeting2rttm))
        if len(meeting_scp) != len(meeting2rttm):
            logging.warning("Number of wav and rttm mismatch {} != {}".format(
                len(meeting_scp), len(meeting2rttm)))
            common_keys = set(meeting_scp.keys()) & set(meeting2rttm.keys())
            logging.warning("Keep {} records.".format(len(common_keys)))
            new_meeting_scp = OrderedDict()
            for key in meeting_scp:
                if key not in common_keys:
                    logging.warning("Pop {} from wav scp".format(key))
                else:
                    new_meeting_scp[key] = meeting_scp[key]
            new_meeting2rttm = OrderedDict()
            for key in meeting2rttm:
                if key not in common_keys:
                    logging.warning("Pop {} from rttm scp".format(key))
                else:
                    new_meeting2rttm[key] = meeting2rttm[key]
            meeting_scp, meeting2rttm = new_meeting_scp, new_meeting2rttm
        if not os.path.exists(args.out_dir):
            os.makedirs(args.out_dir)