diff --git a/metrics_service.sh b/metrics_service.sh index 48edddf..8c5f448 100644 --- a/metrics_service.sh +++ b/metrics_service.sh @@ -12,5 +12,8 @@ awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}' < hypothesis.txt > hypothesis.tr # use sclite to calculate WER, actually we need only lines starting with 'Scores' sclite -f 0 -r reference.trn trn -h hypothesis.trn trn -e utf-8 -i rm -o all stdout | grep "Scores:" > wer_results.txt -cat wer_results.txt | awk 'BEGIN{OFS="\t"}{print ( ($7 + $8 + $9) / ($7 + $8 + &6) ) * 100;}' > wer_per_line.txt +cat wer_results.txt | awk '{print ( ($7 + $8 + $9) / ($7 + $8 + &6) ) * 100;}' >> wer_per_line.txt paste wer_per_line.txt wikiniews_results.tsv + +# calculate mean WER for all records +awk '{sum += &1; n++} END { print sum / n; }' < wer_per_lines.txt > wer.txt