s416178-mlworkshops/shell_script.sh

19 lines
706 B
Bash
Raw Normal View History

2020-04-03 11:00:04 +02:00
#!/bin/bash
2020-04-03 13:18:44 +02:00
wc -l $1 > result.log
2020-04-03 12:44:08 +02:00
cut -f 2 $1 > hypothesis.txt
cut -f 3 $1 > 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
2020-04-03 13:18:44 +02:00
sclite -f 0 -r reference.trn trn -h hypothesis.trn trn -e utf-8 -i rm -o all stdout | grep Scores > scores
2020-04-03 13:25:19 +02:00
awk 'BEGIN{FS=OFS="\t"}{ print $6, $7, $8, $9 }' < scores > step1
cat step1
awk '{ print ($2 + $3 + $4)/($2 + $3 + $1) }' < step1 > wer.tsv
2020-04-03 13:20:13 +02:00
cat wer.tsv
2020-04-03 13:25:19 +02:00
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