Creating a UWP DLL using Windows::Media::SpeechSyn

问题:

I am currently trying to develop a speech synthesis UWP DLL using the namespace Windows::Media::SpeechSynthesis. I read this documentation and the Microsoft page dedicated to the namespace. I tried to implement the namespace in code.

Header file

#pragma once

#include <stdio.h>
#include <string>
#include <iostream>
#include <ppltasks.h>

using namespace Windows::Media::SpeechSynthesis;
using namespace Windows::UI::Xaml::Controls;
using namespace Windows::UI::Xaml::Media;
using namespace Windows::Media::Playback;

namespace SDKTemplate
{
    class TextToSpeechDll
    {
        public:
           __declspec( dllexport ) void ttsInitialize();

        private:
           MediaElement ^media;
    };
}

Cpp file

#include "stdafx.h"
#include "Dll2.h"

using namespace SDKTemplate;
using namespace Platform;
using namespace Concurrency;

void TextToSpeechDll::ttsInitialize()
{
    SpeechSynthesizer ^synth = ref new SpeechSynthesizer();
    // The object for controlling the speech synthesis engine (voice).
    synth = ref new SpeechSynthesizer();
    // The string to speak.
    String^ text = "Hello World";

    // Generate the audio stream from plain text.
    task<SpeechSynthesisStream ^> speakTask = create_task( synth->SynthesizeTextToStreamAsync( text ) );
    speakTask.then( [this, text]( task<SpeechSynthesisStream^> synthesisStreamTask )
    {
        SpeechSynthesisStream ^speechStream = synthesisStreamTask.get();
        // Send the stream to the media object.
        // media === MediaElement XAML object.
        media->AutoPlay = true;
        media->SetSource( speechStream, speechStream->ContentType );
        media->Play();
    } );
}

I can load the DLL file and the function I exported. However, when I try to call the function, I get following error

I tried the example on the Microsoft page but it some how doesn't work and I can't figure out why. I also tested the Windows Universal Samples available on Github which is an UWP app regrouping Text-To-Speech and Speech Recognition.

Has someone experienced a similar issue? Shouldn't I use an XAML element when I don't have an interface?

---

Edit 1

I modified the header file regarding the exportation of the function as suggested by @Peter Torr - MSFT

#pragma once

#include <stdio.h>
#include <string>
#include <iostream>
#include <ppltasks.h>

using namespace Windows::Media::SpeechSynthesis;
using namespace Windows::UI::Xaml::Controls;
using namespace Windows::UI::Xaml::Media;
using namespace Windows::Media::Playback;

namespace SDKTemplate
{
   public ref class TextToSpeechDll sealed
   {
      public:
         void ttsInitialize();

      private:
         MediaElement ^media = ref new MediaElement();
   };
}

However, when I compile, I'm getting a new error on this line

speakTask.then( [this]( task<SpeechSynthesisStream^> synthesisStreamTask )

I researched this error and if I understood it correctly it comes from the importation of the DLL function.

In addition, I call the function like this

_ttsUwpDll->ttsInitialize();

Which brings us here

void NxWindowsTtsUwpDll::ttsInitialize()
{
   int retVal = 0;
   try
   {
      retVal = _ttsInitialize();
   }
   catch( ... )
   {
      printf( "Exception in ttsInitialize\n" );
   }
   //return retVal;
}

回答1:

In MainPage,I initialize the dll file and invoke "ttsInitialize" function like below code.

MainPage::MainPage()
{
 InitializeComponent();
 TextToSpeechDll* gf = new TextToSpeechDll();
 gf->ttsInitialize();
}

And In Dll.h file,I initialize MediaElement like below code,and other code as same as you.

MediaElement^ media = ref new MediaElement();

When I run the project,it works. You can try it and if you still have issues,please show the details that you initialize your dll file.

回答2:

I found an answer to my question. Instead of using MediaElement, I used MediaPlayer. Now it works, but I still need to figure out how to make the engine speak without limiting it in time. The Sleep( 3000 ) means that the voice will speak for 3 seconds. However, if the sentence is longer than 3 seconds, it will be cut. Here is the code of the program.

int TextToSpeechUwpDll::ttsSpeak( const char* text )
{
   SpeechSynthesizer ^speak = ref new SpeechSynthesizer();
   MediaPlayer ^player = ref new MediaPlayer;

   int wchars_num = MultiByteToWideChar( CP_ACP, 0, text, -1, NULL, 0 );
   wchar_t* texts = new wchar_t[wchars_num];
   MultiByteToWideChar( CP_ACP, 0, text, -1, texts, wchars_num );
   String ^sentence = ref new String( texts );

   task<SpeechSynthesisStream ^> speakTask = create_task( speak->SynthesizeTextToStreamAsync( sentence ) );
   speakTask.then( [player, sentence]( SpeechSynthesisStream ^speechStream )
   {
      player->Source = MediaSource::CreateFromStream( speechStream, speechStream->ContentType );
      player->AutoPlay = false;
      player->Play();
      Sleep( 3000 );
   } );

   return true;
}