Discrepancy in Metric Calculation (WER, U-WER, B-WER, Recall) onMALA-ASR with Slidespeech Test Set

[Caesarhhh]

I ran inference using the Slidespeech test set on MALA-ASR and generated thedecode_test beam4_pred and decode test beam4 gt files. However, the metrics(WER, U-WER, B-WER, Recall) I calculated from these files differ significantly from the results reported in the paper. Could you please provide the code used to compute these metrics for the experiments in the paper?

[yanghaha0908]

The code has been uploaded to #208.