s426274-mlworkshops/script.sh
2020-04-22 01:56:52 +02:00

22 lines
1.0 KiB
Bash

wc -l wikiniews_results.tsv > CountResult.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 reference.trn trn -h hypothesis.trn trn -e utf-8 -i rm -o all stdout | grep 'Scores' > wyniki
cat wyniki | tail -n10
cat wyniki | awk 'BEGIN{OFS="\t"}{print $6, $7, $8, $9}' | awk '{ print ($2 + $3 + $4)/($2 + $3 + $1) }' | tail -n 10
cat wyniki | awk 'BEGIN{OFS="\t"}{print $6, $7, $8, $9}' | awk '{ print ($2 + $3 + $4)/($2 + $3 + $1) }' > werAll.txt
awk '{ sum += $1; n++ } END { print sum / n; }' < werAll.txt >> wer.txt
paste wikiniews_results.tsv wer.txt > wikinews_results.tsv
rm wikiniews_results.tsv
calc(){ awk "BEGIN { print "$*" }"; }
all_lines=$( wc -l < werAll.txt )
lines_with_0=$( grep -c '^0$' werAll.txt )
calc $lines_with_0/$all_lines >> srr.txt
tail -n50 wer.txt | sponge wer.txt
tail -n50 srr.txt | sponge srr.txt