dawjur-mlworkshops/calc_metrics.sh

#!/usr/bin/env bash
cut -f2 wikiniews_results.tsv | awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}' > hyp.trn
cut -f3 wikiniews_results.tsv | awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}' > ref.trn
sclite -f 0 -r ref.trn trn -h hyp.trn trn -e utf-8 -i rm -o sum stdout | grep "Sum/Avg" | python -c "import sys; import re; print('\t'.join(re.findall(r'(\d+\.\d+)', sys.stdin.read())[-2:]))" > wer_srr.txt
cut -f1 wer_srr.txt > wer_total.txt
cut -f2 wer_srr.txt >> srr.txt

sclite -f 0 -r ref.trn trn -h hyp.trn trn -e utf-8 -i rm -o pra stdout | grep "Scores:" | sed 's/Scores: (#C #S #D #I) // ' | awk '{ c=$1; s=$2; d=$3; i=$4; print ($s+$d+$i)/($s+$d+$c)}' > wer_all.txt
awk '{ sum += $1; n++ } END { print sum / n; }' < wer_all.txt >> wer.txt

paste wikiniews_results.tsv wer.txt > wikinews_results.tsv
rm wikiniews_results.tsv

tail -n20 wer.txt | sponge wer.txt
tail -n20 srr.txt | sponge srr.txt