Fix voice input for iOS Safari and prevent infinite loop

- Remove temperature parameter from GPT-5-mini activity extraction (not supported)
- Add classification state to useVoiceInput hook to avoid duplicate API calls
- Prevent infinite loop in VoiceFloatingButton by tracking lastClassifiedTranscript
- Use classification from backend directly instead of making second request
- iOS Safari now successfully transcribes with Azure Whisper and classifies with GPT-5-mini

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

maternal-web/app/api/voice/transcribe/route.ts

@@ -30,16 +30,52 @@ export async function POST(request: NextRequest) {
         );
       }
     } else if (contentType.includes('multipart/form-data')) {
-      // Audio file upload (needs transcription)
-      // TODO: Implement Whisper API integration for audio transcription
-      // For now, return not implemented
+      // Audio file upload - forward to backend for Whisper transcription
+      const formData = await request.formData();
+      const audioFile = formData.get('audio');
+
+      if (!audioFile) {
+        return NextResponse.json(
+          {
+            error: 'VOICE_NO_AUDIO',
+            message: 'No audio file provided',
+          },
+          { status: 400 }
+        );
+      }
+
+      // Forward to backend
+      const backendUrl = process.env.NEXT_PUBLIC_API_URL || 'http://localhost:3020';
+      const backendFormData = new FormData();
+      backendFormData.append('audio', audioFile);
+
+      const backendResponse = await fetch(`${backendUrl}/api/v1/voice/transcribe`, {
+        method: 'POST',
+        body: backendFormData,
+        headers: {
+          // Forward auth token if present
+          ...(request.headers.get('authorization') && {
+            authorization: request.headers.get('authorization')!,
+          }),
+        },
+      });
+
+      if (!backendResponse.ok) {
+        const errorData = await backendResponse.json();
+        return NextResponse.json(errorData, { status: backendResponse.status });
+      }
+
+      const result = await backendResponse.json();
+
+      // Backend returns { success, transcript, classification }
+      // Return in the format expected by the frontend
       return NextResponse.json(
         {
-          error: 'VOICE_AUDIO_NOT_IMPLEMENTED',
-          message: 'Audio transcription not yet implemented. Use text input for now.',
-          hint: 'Send JSON with { "text": "your voice command" }',
+          success: true,
+          transcript: result.transcript,
+          classification: result.classification,
         },
-        { status: 501 }
+        { status: 200 }
       );
     } else {
       return NextResponse.json(