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