I'm writing a plugin (dll file) for speech recognition, and I'm creating a WinForm as its interface/dialog. When I run the plugin and click the 'Speak' to start the initialization, I get an unhandled exception.

Here is a piece of the code:

    public ref class Dialog : public System::Windows::Forms::Form
    {
       public: SpeechRecognitionEngine^ sre;

       private: System::Void btnSpeak_Click(System::Object^  sender, System::EventArgs^  e) 
       {
         Initialize();
       }

       protected: void Initialize()
       {  
          if (System::Threading::Thread::CurrentThread->GetApartmentState() !=
                System::Threading::ApartmentState::STA)
          {
              throw gcnew InvalidOperationException("UI thread required");
          }

          //create the recognition engine
          sre = gcnew SpeechRecognitionEngine();

          //set our recognition engine to use the default audio device
          sre->SetInputToDefaultAudioDevice();

          //create a new GrammarBuilder to specify which commands we want to use
          GrammarBuilder^ grammarBuilder = gcnew GrammarBuilder();

          //append all the choices we want for commands.
          //we want to be able to move, stop, quit the game, and check for the cake.
          grammarBuilder->Append(gcnew Choices("play", "stop"));

          //create the Grammar from th GrammarBuilder
          Grammar^ customGrammar = gcnew Grammar(grammarBuilder);

          //unload any grammars from the recognition engine
          sre->UnloadAllGrammars();

          //load our new Grammar
          sre->LoadGrammar(customGrammar);

          //add an event handler so we get events whenever the engine recognizes spoken commands
          sre->SpeechRecognized += gcnew EventHandler<SpeechRecognizedEventArgs^> (this, &Dialog::sre_SpeechRecognized);

          //set the recognition engine to keep running after recognizing a command.
              //if we had used RecognizeMode.Single, the engine would quite listening after
          //the first recognized command.
          sre->RecognizeAsync(RecognizeMode::Multiple);

          //this->init();
       }  

       void sre_SpeechRecognized(Object^ sender, SpeechRecognizedEventArgs^ e)
       {
          //simple check to see what the result of the recognition was
          if (e->Result->Text == "play")
          {
             MessageBox(plugin.hwndParent, L"play", 0, 0);
          }

                  if (e->Result->Text == "stop")
          {
             MessageBox(plugin.hwndParent, L"stop", 0, 0);
          }
       }
    };