s460941-metrics/count.sh
2020-04-23 21:33:03 +02:00

15 lines
724 B
Bash
Executable File

var=$(wc -l "wikiniews_results.tsv")
lines=${var%% *}
echo $lines > "lines_number.txt"
cut -f2 wikiniews_results.tsv > hypothesis.txt
cut -f3 wikiniews_results.tsv > reference.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 ref.trn trn -h hyp.trn trn -e utf-8 -i rm -o all stdout | grep "Scores:" | sed 's/Scores: (#C #S #D #I) // ' | awk '{ print ($2+$3+$4)/($1+$2+$3)}' > wer_results.txt
awk '{ all+=$0;count++ } END { print all/count }' wer_results.txt > wer.txt
awk ' $1==0 {score[0]++} $1!=0 {score[1]++} END { s=score[0]+score[1]; print(score[0]/s)*100}' wer_results.txt > srr.txt