fix(annotation): converters doesn't read segment labels correctly

IPS-LMU · Sep 19, 2023 · 5d8ce29 · 5d8ce29
1 parent 3535df3
commit 5d8ce29
Show file tree

Hide file tree

Showing 12 changed files with 37 additions and 23 deletions.
diff --git a/libs/annotation/package.json b/libs/annotation/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@octra/annotation",
-  "version": "0.1.3",
+  "version": "0.1.4",
   "type": "module",
   "peerDependencies": {
     "x2js": "^3.4.4"

diff --git a/libs/annotation/project.json b/libs/annotation/project.json
@@ -6,7 +6,6 @@
   "targets": {
     "build": {
       "executor": "@nrwl/rollup:rollup",
-      "dependsOn": ["media"],
       "outputs": [
         "{options.outputPath}"
       ],

diff --git a/libs/annotation/src/lib/annotjson.ts b/libs/annotation/src/lib/annotjson.ts
@@ -126,7 +126,7 @@ export class OAnnotJSON
   }
 
   static deserialize(jsonObject: IAnnotJSON): OAnnotJSON | undefined {
-    if(jsonObject) {
+    if (jsonObject) {
       return new OAnnotJSON(
         jsonObject.annotates,
         jsonObject.name,
@@ -135,7 +135,7 @@ export class OAnnotJSON
         jsonObject.links
       );
     }
-    return undefined
+    return undefined;
   }
 
   deserialize(jsonObject: IAnnotJSON): OAnnotJSON | undefined {
@@ -155,7 +155,11 @@ export class OLevel<T extends OItem> implements ILevel {
   }
 
   clone() {
-    return new OLevel(this.name, this.type, this.items.map(a => a.clone()));
+    return new OLevel(
+      this.name,
+      this.type,
+      this.items.map((a) => a.clone())
+    );
   }
 }
 
@@ -280,6 +284,10 @@ export class OItem implements IItem, Serializable<IItem, OItem> {
     return new OItem(jsonObject.id, jsonObject.labels);
   }
 
+  getFirstLabelWithoutName(notName: string) {
+    return this.labels?.find((a) => a.name !== notName);
+  }
+
   clone(id?: number) {
     return new OItem(id ?? this.id, [...this.labels]);
   }
@@ -413,7 +421,7 @@ export class OLink implements ILink, Serializable<ILink, OLink> {
     return new OLink(jsonObject.fromID, jsonObject.toID);
   }
 
-  clone(){
+  clone() {
     return new OLink(this.fromID, this.toID);
   }
 }

diff --git a/libs/annotation/src/lib/converters/CTMConverter.ts b/libs/annotation/src/lib/converters/CTMConverter.ts
@@ -63,7 +63,8 @@ export class CTMConverter extends Converter {
     const level = annotation.levels[levelnum];
 
     for (const levelItem of level.items as OSegment[]) {
-      const transcript = levelItem.labels[0].value;
+      const transcript =
+        levelItem.getFirstLabelWithoutName('Speaker')?.value ?? '';
       const start =
         Math.round((levelItem.sampleStart! / audiofile.sampleRate) * 100) / 100;
       const duration =

diff --git a/libs/annotation/src/lib/converters/ELANConverter.ts b/libs/annotation/src/lib/converters/ELANConverter.ts
@@ -111,7 +111,7 @@ export class ELANConverter extends Converter {
             ALIGNABLE_ANNOTATION: {
               _ANNOTATION_ID: `a${aidCounter}`,
               ANNOTATION_VALUE:
-                segment.labels.find((a) => a.name !== 'Speaker')?.value ?? '',
+                segment.getFirstLabelWithoutName('Speaker')?.value ?? '',
               _TIME_SLOT_REF1: `ts${tsidCounter - 1}`,
               _TIME_SLOT_REF2: `ts${tsidCounter}`,
             },

diff --git a/libs/annotation/src/lib/converters/PartiturConverter.ts b/libs/annotation/src/lib/converters/PartiturConverter.ts
@@ -70,9 +70,9 @@ LBD:\n`;
       let ortCounter = 0;
 
       for (const item of annotation.levels[levelnum].items as OSegment[]) {
-        const words = item.labels
-          .filter((a) => a.name !== 'Speaker')[0]
-          .value.split(' ');
+        const words = (
+          item.getFirstLabelWithoutName('Speaker')?.value ?? ''
+        ).split(' ');
         ort = ort.concat(words);
         let trnLine = `TRN: ${item.sampleStart} ${item.sampleDur} `;
 
@@ -84,7 +84,7 @@ LBD:\n`;
         }
         ortCounter += words.length;
         trnLine += ` ${
-          item.labels.filter((a) => a.name !== 'Speaker')[0].value
+          item.getFirstLabelWithoutName('Speaker')?.value ?? ''
         }\n`;
         trn.push(trnLine);
       }

diff --git a/libs/annotation/src/lib/converters/PraatTableConverter.ts b/libs/annotation/src/lib/converters/PraatTableConverter.ts
@@ -6,7 +6,6 @@ import {
   OctraAnnotationFormatType,
 } from './Converter';
 import {
-  ISegment,
   OAnnotJSON,
   OAnyLevel,
   OLabel,
@@ -47,14 +46,13 @@ export class PraatTableConverter extends Converter {
     const addEntry = (
       res: string,
       level: OAnyLevel<OSegment>,
-      segment: ISegment
+      segment: OSegment
     ) => {
       const tmin = segment.sampleStart / annotation.sampleRate;
       const tmax =
         (segment.sampleStart + segment.sampleDur) / annotation.sampleRate;
-      const transcript = segment.labels.find(
-        (a) => a.name !== 'Speaker'
-      )?.value;
+      const transcript =
+        segment.getFirstLabelWithoutName('Speaker')?.value ?? '';
 
       return `${res}${tmin}\t${level.name}\t${transcript}\t${tmax}\n`;
     };
@@ -64,8 +62,8 @@ export class PraatTableConverter extends Converter {
     for (const level of annotation.levels) {
       // export segments only
       if (level.type === 'SEGMENT') {
-        for (const segment of level.items) {
-          result = addEntry(result, level, segment as ISegment);
+        for (const segment of level.items as OSegment[]) {
+          result = addEntry(result, level, segment);
         }
       }
     }

diff --git a/libs/annotation/src/lib/converters/PraatTextgridConverter.ts b/libs/annotation/src/lib/converters/PraatTextgridConverter.ts
@@ -99,7 +99,7 @@ export class PraatTextgridConverter extends Converter {
             `            xmin = ${secondsStart} \n` +
             `            xmax = ${secondsEnd} \n` +
             `            text = "${
-              segment.labels.find((a) => a.name !== 'Speaker')?.value
+              segment.getFirstLabelWithoutName('Speaker')?.value ?? ''
             }" \n`;
         }
       }

diff --git a/libs/annotation/src/lib/converters/SRTConverter.ts b/libs/annotation/src/lib/converters/SRTConverter.ts
@@ -86,7 +86,8 @@ export class SRTConverter extends Converter {
       let counter = 1;
       if (level.type === 'SEGMENT') {
         for (const item of level.items as OSegment[]) {
-          const transcript = item.labels[0].value;
+          const transcript =
+            item.getFirstLabelWithoutName('Speaker')?.value ?? '';
           const start = this.getTimeStringFromSamples(
             item.sampleStart,
             annotation.sampleRate

diff --git a/libs/annotation/src/lib/converters/TextConverter.ts b/libs/annotation/src/lib/converters/TextConverter.ts
@@ -65,7 +65,8 @@ export class TextConverter extends Converter {
       if (level.type === 'SEGMENT') {
         for (let j = 0; j < level.items.length; j++) {
           const item = level.items[j] as OSegment;
-          const transcript = item.labels[0].value;
+          const transcript =
+            item.getFirstLabelWithoutName('Speaker')?.value ?? '';
 
           result += transcript;
           if (j < level.items.length - 1) {

diff --git a/libs/annotation/src/lib/converters/WebVTTConverter.ts b/libs/annotation/src/lib/converters/WebVTTConverter.ts
@@ -63,7 +63,9 @@ export class WebVTTConverter extends Converter {
       if (level.type === 'SEGMENT') {
         for (let j = 0; j < level.items.length; j++) {
           const item = level.items[j] as OSegment;
-          const transcript = item.labels[0].value
+          const transcript = (
+            item.getFirstLabelWithoutName('Speaker')?.value ?? ''
+          )
             .replace(/</g, '&lt;')
             .replace(/>/g, '&gt;');
           const start = this.getTimeStringFromSamples(

diff --git a/libs/annotation/src/lib/octraAnnotationSegment.ts b/libs/annotation/src/lib/octraAnnotationSegment.ts
@@ -49,6 +49,10 @@ export class OctraAnnotationEvent
       ...this.labels,
     ]);
   }
+
+  getFirstLabelWithoutName(notName: string) {
+    return this.labels?.find((a) => a.name !== notName);
+  }
 }
 
 export class OctraAnnotationSegment<T extends ASRContext = ASRContext>