s442676-mlworkshops/metryki.sh

15 lines
743 B
Bash

cut -f 2 wikiniews_results.tsv > reference.txt
cut -f 3 wikiniews_results.tsv > hypothesis.txt
awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}' < hypothesis.txt > hypothesis.trn
awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}' < reference.txt > reference.trn
sclite -f 0 -r reference.trn trn -h hypothesis.trn trn -e utf-8 -i rm -o all stdout > wynik.txt
grep 'Scores:' wynik.txt | awk '{print $6, $7, $8, $9}' > scores.txt
awk '{ print ($2 + $3 + $4) / ($2 + $3 + $1) }' scores.txt > WER.txt
awk '{ n++; if ($2 == 0 && $3 == 0 && $4 == 0) poprawne += 1 } END { print poprawne/n }' scores.txt > SSR.txt
awk '{ n++; suma += $1 } END { print suma/n }' WER.txt > sredniaWER.txt
paste wikiniews_results.tsv WER.txt > wikiniews_results_WER.tsv