feat(stt): pass last assistant response as Whisper context to reduce mishearings

lib/data/api/voice_api.dart

@@ -40,16 +40,20 @@ class VoiceApi {
   }
 
   /// POST WebM/Opus audio bytes and return the transcript string.
+  /// [context] is optional recent conversation text passed as initial_prompt
+  /// to Whisper, reducing mishearings of domain-specific words.
   /// Returns empty string on empty or error response.
-  Future<String> transcribe(Uint8List audioBytes) async {
+  Future<String> transcribe(Uint8List audioBytes, {String? context}) async {
     try {
-      final formData = FormData.fromMap({
+      final fields = <String, dynamic>{
         'audio': MultipartFile.fromBytes(
           audioBytes,
           filename: 'audio.m4a',
           contentType: DioMediaType('audio', 'mp4'),
         ),
-      });
+        if (context != null && context.isNotEmpty) 'context': context,
+      };
+      final formData = FormData.fromMap(fields);
       final response = await _dio.post(
         '/api/voice/transcribe',
         data: formData,

lib/data/repositories/voice_repository.dart

@@ -7,6 +7,7 @@ class VoiceRepository {
   const VoiceRepository(this._api);
 
   Future<VoiceStatus> checkStatus() => _api.checkStatus();
-  Future<String> transcribe(Uint8List audioBytes) => _api.transcribe(audioBytes);
+  Future<String> transcribe(Uint8List audioBytes, {String? context}) =>
+      _api.transcribe(audioBytes, context: context);
   Future<Uint8List> synthesise(String text) => _api.synthesise(text);
 }

lib/providers/voice_provider.dart

@@ -108,6 +108,10 @@ class VoiceNotifier extends Notifier<VoiceState> {
   int _lastSeenLength = 0;
   bool _streamComplete = false;
 
+  // Last complete assistant response — passed to Whisper as initial_prompt
+  // to reduce STT mishearings of domain-specific words.
+  String _lastAssistantContent = '';
+
   // TTS playback queue
   final _ttsQueue = Queue<Uint8List>();
   bool _ttsPlaying = false;
@@ -205,6 +209,7 @@ class VoiceNotifier extends Notifier<VoiceState> {
     _dispatchSentences(flush: isComplete);
 
     if (isComplete) {
+      _lastAssistantContent = fullContent;
       _streamComplete = true;
       _checkRestartListening();
     }
@@ -278,8 +283,10 @@ class VoiceNotifier extends Notifier<VoiceState> {
 
       if (!state.voiceModeActive) return;
 
-      final transcript =
-          await ref.read(voiceRepositoryProvider).transcribe(bytes);
+      final transcript = await ref.read(voiceRepositoryProvider).transcribe(
+            bytes,
+            context: _lastAssistantContent.isNotEmpty ? _lastAssistantContent : null,
+          );
 
       if (!state.voiceModeActive) return;