grep WER

Jenkinsfile

@@ -34,8 +34,9 @@ pipeline {
                 sh './zadanie3/trn.sh'
                 archiveArtifacts 'zadanie3/hypothesis.trn'
                 archiveArtifacts 'zadanie3/reference.trn'
-                archiveArtifacts 'zadanie3/results.txt'
+
-                archiveArtifacts 'zadanie3/results-p.txt'
+                sh 'chmod 755 ./zadanie3/grep.sh'
+                sh './zadanie3/grep.sh'
             }
         }
     }

zadanie3/grep.sh

@@ -0,0 +1,7 @@
+
+cat zadanie3/sclite.txt | grep -oP '(?<=Scores: \(#C #S #D #I\) ).*' > results_wer.txt
+
+# (S + D + I)/(S + D + C)
+awk '{ print ($2 + $3 + $4) / ($2 + $3 + $1) }' results_wer.txt
+
+cat results_wer.txt

zadanie3/script.py

@@ -1,11 +0,0 @@
-#!/usr/bin/env python3
-# -*- coding: utf-8 -*-
-
-import numpy, pandas
-
-input_data = pandas.read_csv('ASR-eval/wikiniews_results.tsv', encoding='latin1', error_bad_lines=False, sep='\t', header=None)
-sentences = input_data[1]
-references = input_data[2]
-
-numpy.savetxt('zadanie3/sentences.txt', sentences, '%s')
-numpy.savetxt('zadanie3/references.txt', references, '%s')

zadanie3/trn.sh

@@ -5,5 +5,4 @@ cut -f 3 ASR-eval/wikiniews_results.tsv > zadanie3/reference.txt
 awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}'  < zadanie3/hypothesis.txt > zadanie3/hypothesis.trn
 awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}'  < zadanie3/reference.txt > zadanie3/reference.trn
 
-sclite -f 0 -r zadanie3/reference.trn trn -h zadanie3/hypothesis.trn trn -e utf-8 -i rm -o all stdout > zadanie3/results.txt
+sclite -f 0 -r zadanie3/reference.trn trn -h zadanie3/hypothesis.trn trn -e utf-8 -i rm -o all stdout > zadanie3/sclite.txt
-sclite -f 0 -r zadanie3/reference.trn trn -h zadanie3/hypothesis.trn trn -e utf-8 -i rm -o all -p > zadanie3/results-p.txt