chore: save voice in TTS events and re-request if requested voice doesn't match saved voice

lib/pangea/common/constants/model_keys.dart

@@ -109,6 +109,7 @@ class ModelKey {
 
   static const String transcription = "transcription";
   static const String botTranscription = 'bot_transcription';
+  static const String voice = "voice";
 
   // bot options
   static const String languageLevel = "difficulty";

lib/pangea/events/event_wrappers/pangea_message_event.dart

@@ -303,7 +303,11 @@ class PangeaMessageEvent {
     return null;
   }
 
-  Event? getTextToSpeechLocal(String langCode, String text) {
+  Event? getTextToSpeechLocal(
+    String langCode,
+    String text,
+    String? voice,
+  ) {
     for (final audio in allAudio) {
       final dataMap = audio.content.tryGetMap(ModelKey.transcription);
       if (dataMap == null || !dataMap.containsKey(ModelKey.tokens)) continue;
@@ -313,7 +317,9 @@ class PangeaMessageEvent {
           dataMap as dynamic,
         );
 
-        if (audioData.langCode == langCode && audioData.text == text) {
+        if (audioData.langCode == langCode &&
+            audioData.text == text &&
+            audioData.voice == voice) {
           return audio;
         }
       } catch (e, s) {
@@ -368,7 +374,7 @@ class PangeaMessageEvent {
     String langCode,
     String? voice,
   ) async {
-    final local = getTextToSpeechLocal(langCode, messageDisplayText);
+    final local = getTextToSpeechLocal(langCode, messageDisplayText, voice);
     if (local != null) {
       final file = await local.getPangeaAudioFile();
       if (file != null) return file;
@@ -424,7 +430,7 @@ class PangeaMessageEvent {
           'waveform': response.waveform,
         },
         ModelKey.transcription: response
-            .toPangeaAudioEventData(rep?.text ?? body, langCode)
+            .toPangeaAudioEventData(rep?.text ?? body, langCode, voice)
             .toJson(),
       },
     ).then((eventId) async {

lib/pangea/text_to_speech/text_to_speech_response_model.dart

@@ -41,11 +41,16 @@ class TextToSpeechResponseModel {
         "tts_tokens": List<dynamic>.from(ttsTokens.map((x) => x.toJson())),
       };
 
-  PangeaAudioEventData toPangeaAudioEventData(String text, String langCode) {
+  PangeaAudioEventData toPangeaAudioEventData(
+    String text,
+    String langCode,
+    String? voice,
+  ) {
     return PangeaAudioEventData(
       text: text,
       langCode: langCode,
       tokens: ttsTokens,
+      voice: voice,
     );
   }
 }
@@ -91,11 +96,13 @@ class PangeaAudioEventData {
   final String text;
   final String langCode;
   final List<TTSToken> tokens;
+  final String? voice;
 
   PangeaAudioEventData({
     required this.text,
     required this.langCode,
     required this.tokens,
+    this.voice,
   });
 
   factory PangeaAudioEventData.fromJson(dynamic json) => PangeaAudioEventData(
@@ -106,6 +113,7 @@ class PangeaAudioEventData {
               .map((x) => TTSToken.fromJson(x))
               .toList(),
         ),
+        voice: json[ModelKey.voice] as String?,
       );
 
   Map<String, dynamic> toJson() => {
@@ -113,5 +121,6 @@ class PangeaAudioEventData {
         ModelKey.langCode: langCode,
         ModelKey.tokens:
             List<Map<String, dynamic>>.from(tokens.map((x) => x.toJson())),
+        if (voice != null) ModelKey.voice: voice,
       };
 }