s460941-metrics/count.sh

15 lines
739 B
Bash
Raw Normal View History

2020-04-22 16:44:59 +02:00
var=$(wc -l "wikiniews_results.tsv")
lines=${var%% *}
2020-04-22 16:55:50 +02:00
echo $lines > "lines_number.txt"
2020-04-23 20:42:39 +02:00
2020-04-23 22:32:26 +02:00
cut -f2 wikiniews_results.tsv > hypothesis.txt
cut -f3 wikiniews_results.tsv > reference.txt
2020-04-23 21:59:51 +02:00
2020-04-23 22:32:26 +02:00
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
2020-04-23 21:59:51 +02:00
2020-04-23 22:38:29 +02:00
sclite -f 0 -r reference.trn trn -h hypothesis.trn trn -e utf-8 -i rm -o all stdout | grep Scores | awk -v OFS='\t' '{print $6,$7,$8,$9}' | awk '{print($2+$3+$4)/($1+$2+$3) }' > wer_results.txt
2020-04-23 20:42:39 +02:00
2020-04-23 23:11:16 +02:00
awk '{ sum+=$0;count++ } END { print sum/count }' wer_results.txt >> wer.txt
2020-04-23 22:53:08 +02:00
awk ' $0==0 {score[0]++} $0!=0 {score[1]++} END { s=score[0]+score[1]; print(score[0]/s)*100}' wer_results.txt >> srr.txt