s416226-mlworkshops/script.sh

14 lines
742 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_for_all.txt
awk '{ total += $0; count++ } END { print total/count }' wer_for_all.txt > wer.txt
awk ' $1 == 0 { temp[0]++ } $1 != 0 { temp[1]++ }END { t = temp[0] + temp[1]; print (temp[0]/t) * 100}' wer_for_all.txt > srr.txt