Add scripts for generating wer and srr

Dockerfile

@@ -3,6 +3,7 @@ FROM ubuntu:latest
 RUN apt update -y && apt install -y make
 RUN apt install -y git
 RUN apt install -y gcc
+RUN apt install -y python3
 RUN gcc --version
 RUN apt install -y build-essential
 RUN git clone https://github.com/usnistgov/SCTK.git

Jenkinsfile

@@ -18,7 +18,7 @@ pipeline {
         }
         stage('archiveArtifacts') {
             steps {
-                archiveArtifacts 'results.txt'
+                archiveArtifacts 'results.txt, wer.txt, srr.txt'
             }
         }
     }

count_metrics.py

@@ -0,0 +1,26 @@
+import sys
+
+if __name__ == '__main__':
+    word_error_rate = 0
+    number_of_sentence = 0
+    correct_sentences = 0
+
+    with open(sys.argv[1], 'r') as sclite_result, \
+            open('wer.txt', 'w') as wer, \
+            open('srr.txt', 'w') as srr:
+
+        for line in sclite_result:
+            complited = False
+            if line[:7] == 'Scores:':
+                number_of_sentence += 1
+                c, s, d, i = line.strip().split()[-4:]
+                word_error_rate = sum(map(int, [s, d, i])) / sum(map(int, [s, d, c]))
+                complited = True
+
+            if complited:
+                wer.write(f'{word_error_rate:.5f}\n')
+                if word_error_rate == 0:
+                    correct_sentences += 1
+
+        sentence_recognition_rate = correct_sentences / number_of_sentence
+        srr.write(f'{sentence_recognition_rate:.5f}\n')

script.sh

@@ -4,3 +4,5 @@ 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
 
 sclite -f 0 -r reference.trn trn -h hypothesis.trn trn -e utf-8 -i rm -o all stdout > results.txt
+python3 count_metrics.py results.txt
+paste results.txt wer.txt > tmp.txt && mv tmp.txt results.txt