diff --git a/script.sh b/script.sh index 51ec39a..1c83ee2 100644 --- a/script.sh +++ b/script.sh @@ -1,15 +1,13 @@ wc -l wikiniews_results.tsv > res.txt -awk -F "\t" '{print $3}' wikiniews_results.tsv > reference.txt -awk -F "\t" '{print $2}' wikiniews_results.tsv > hypothesis.txt +# create trn +cut -f2 $1 | awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}' > hypothesis.trn +cut -f3 $1 | awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}' > reference.trn -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 +#wer for each line +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)/($2 + $3 + $1) }' > wer_per_line.tsv +paste $1 wer_per_line.tsv > wikiniews_results_with_wer.tsv -sclite -f 0 -r reference.trn trn -h hypothesis.trn trn -e utf-8 -i rm -o all stdout - -echo 'hypo' -cat hypothesis.trn - -echo 'refe' -cat reference.trn \ No newline at end of file +#average wer and srr +awk -F'\t' '{ total += $5; count++ } END { print total/count }' wikiniews_results_with_wer.tsv > wer.txt +awk -F'\t' '{ if ( $5 == 0 ) good++; count++ } END { print good/count }' wikiniews_results_with_wer.tsv > srr.txt \ No newline at end of file