s416226-mlworkshops/script.sh

12 lines
555 B
Bash

wc -l wikiniews_results.tsv > lines_number.txt
cut -f 2 wikiniews_results.tsv > hypothesis.txt
cut -f 3 wikiniews_results.tsv > reference.txt
awk 'BEGIN{OFS="\t"}{print $0,"(sp1_"NR")"}' < hypothesis.txt > hypothesis.trn
awk 'BEGIN{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 | grep Scores > Scores
awk 'BEGIN{OFS="\t"}{print $6, $7, $8, $9}' < Scores | awk '{ print ($2 + $3 + $4)/($2 + $3 + $1) }' > wer.txt
grep "^0$" Scores | wc -l > srr.txt