Cleanup texts for calculating WER

Florian Treml · Florian Treml · commit c3afd21a3a33 · 2021-12-30T14:44:19.000+01:00
diff --git a/frontend/src/utils.js b/frontend/src/utils.js
@@ -23,10 +23,14 @@ const enumValueToName = (obj, value) => {
   return Object.keys(obj)[Object.values(obj).findIndex(x => x === value)]
 }
 
+const _cleanString = (str) => str.toLowerCase().replace(/[\W_]+/g, ' ').split(' ').filter(s => s && s.length > 0).join(' ')
+
 const wer = async (text1, text2) => {
+  const from = _cleanString(text1 || '')
+  const to = _cleanString(text2 || '')
   return {
-    distance: speechScorer.calculateEditDistance(text1 || '', text2 || ''),
-    wer: speechScorer.wordErrorRate(text1 || '', text2 || '')
+    distance: speechScorer.calculateEditDistance(from, to),
+    wer: speechScorer.wordErrorRate(from, to)
   }
 }
 

Original file line number	Diff line number	Diff line change
`@@ -23,10 +23,14 @@ const enumValueToName = (obj, value) => {`
`23`	`23`	`return Object.keys(obj)[Object.values(obj).findIndex(x => x === value)]`
`24`	`24`	`}`
`25`	`25`
	`26`	`+const _cleanString = (str) => str.toLowerCase().replace(/[\W_]+/g, ' ').split(' ').filter(s => s && s.length > 0).join(' ')`
	`27`	`+`
`26`	`28`	`const wer = async (text1, text2) => {`
	`29`	`+ const from = _cleanString(text1 \|\| '')`
	`30`	`+ const to = _cleanString(text2 \|\| '')`
`27`	`31`	`return {`
`28`		`- distance: speechScorer.calculateEditDistance(text1 \|\| '', text2 \|\| ''),`
`29`		`- wer: speechScorer.wordErrorRate(text1 \|\| '', text2 \|\| '')`
	`32`	`+ distance: speechScorer.calculateEditDistance(from, to),`
	`33`	`+ wer: speechScorer.wordErrorRate(from, to)`
`30`	`34`	`}`
`31`	`35`	`}`
`32`	`36`