s416267-mlworkshops/lines.sh
Kamil Paprota 902e222040 metryki
2020-04-03 15:43:06 +02:00

16 lines
808 B
Bash

wc -l wikiniews_results.tsv > 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")"}' < 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{OFS="\t"}{print $6, $7, $8, $9}' < scores | awk '{ print ($2 + $3 + $4)/($2 + $3 + $1) }' > wer.tsv
awk '{getline l < "wer.trn"; print $0"\t"l} ' wikiniews_results.tsv > wikiniews_results.tsv
awk 'BEGIN{FS=OFS="\t"}{ bad = bad + $1; all++ }END{ print bad/all }' < wer.tsv >> wer.txt
awk 'BEGIN{FS=OFS="\t"}{ if ( $1 == 0 ) good++; all++ }END{ print good/all }' < wer.tsv >> srr.txt