使用SpeechLib生成音频时Word事件未触发，求问题排查与解决

Hey there, let's dig into why your Word event isn't firing. I've worked with SpeechLib before, so here are some practical fixes and explanations to get things working:

Possible Issues & Solutions #

1. Adjust the Order of Event Configuration #

In your code, you set v.EventInterests before assigning the AudioOutputStream. The Speech API sometimes requires event settings to be applied after configuring the output stream. Let's tweak that order:

using SpeechLib;
using System;

class Program {
    static void Main(string[] args) {
        SpeakToFile(Console.ReadLine());
    }

    static void SpeakToFile(string toSpeak) {
        SpVoice v = new SpVoice();
        SpFileStream speechStream = new SpFileStream();
        speechStream.Open("D:/Documenten/test.mp3", SpeechStreamFileMode.SSFMCreateForWrite, true);
        
        // Set output stream first
        v.AudioOutputStream = speechStream;
        // Then configure events
        v.EventInterests = SpeechVoiceEvents.SVEAllEvents;
        v.Word += V_Word;
        
        v.Rate = 0;
        v.Speak(toSpeak, SpeechVoiceSpeakFlags.SVSFDefault);
        speechStream.Close();
        
        Console.WriteLine("Press any key to exit...");
        Console.ReadKey();
    }

    static void V_Word(int StreamNumber, object StreamPosition, int CharacterPosition, int Length) {
        Console.WriteLine(CharacterPosition);
    }
}

2. Switch to Asynchronous Speech #

Synchronous speech (SVSFDefault) might not fire events reliably when writing to a file. Try using asynchronous mode and wait for the speech to complete with an EndStream handler:

First, add a completion signal and handler:

using System.Threading;

static ManualResetEvent _speechFinished = new ManualResetEvent(false);

static void V_EndStream(int StreamNumber, object StreamPosition) {
    _speechFinished.Set();
}

Then update the SpeakToFile method:

static void SpeakToFile(string toSpeak) {
    _speechFinished.Reset();
    SpVoice v = new SpVoice();
    SpFileStream speechStream = new SpFileStream();
    speechStream.Open("D:/Documenten/test.mp3", SpeechStreamFileMode.SSFMCreateForWrite, true);
    
    v.AudioOutputStream = speechStream;
    v.EventInterests = SpeechVoiceEvents.SVEAllEvents;
    v.Word += V_Word;
    v.EndStream += V_EndStream; // Attach completion handler
    
    v.Rate = 0;
    // Use asynchronous speech flag
    v.Speak(toSpeak, SpeechVoiceSpeakFlags.SVSFlagsAsync);
    // Wait until speech completes
    _speechFinished.WaitOne();
    
    speechStream.Close();
    Console.WriteLine("Press any key to exit...");
    Console.ReadKey();
}

3. Test Without File Output First #

To rule out file-specific limitations, temporarily remove the AudioOutputStream assignment and let speech play through your default audio device. If the Word event fires then, your version of the Speech API likely doesn't support events when writing to a file. In that case:

• You could pre-process the input text to split into words and track positions manually (imperfect, but workable)
• Ensure you have the latest version of Windows Speech API installed

4. Verify COM Interop Setup #

Double-check your project references:

• If using NuGet, confirm you have the correct SpeechLib interop package installed
• If adding a COM reference, select "Microsoft Speech Object Library" from the COM tab in your project settings

Why This Happens #

SpeechLib is a COM-based library, and .NET interop can be finicky with event timing. The order of configuration or speech mode (synchronous vs asynchronous) often impacts event firing. Additionally, some API versions prioritize file-writing efficiency over event propagation, which can suppress the Word event when saving to a file.

内容的提问来源于stack exchange，提问作者hap