dawjur-mlworkshops/calc_metrics.sh

#!/usr/bin/env bash
cut -f2 wikiniews_results.tsv | awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}' > hyp.trn
cut -f3 wikiniews_results.tsv | awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}' > ref.trn

sclite -f 0 -r ref.trn trn -h hyp.trn trn -e utf-8 -i rm -o pra stdout | grep "Scores:" | sed 's/Scores: (#C #S #D #I) // ' | awk '{ print ($2+$3+$4)/($1+$2+$3)}' > wer_all.txt
awk '{ sum += $1; n++ } END { print sum / n; }' < wer_all.txt >> wer.txt

python3 -c "print(`grep -c '^0$' wer_all.txt`/`cat wer_all.txt | wc -l`)" >> srr.txt
paste wikiniews_results.tsv wer.txt > wikinews_results.tsv
rm wikiniews_results.tsv

rm wer.txt srr.txt
tail -n50 wer.txt | sponge wer.txt
tail -n50 srr.txt | sponge srr.txt