如何在 C# 中使用 SpInprocRecoContext 识别语音事件？

Posted 2023-03-10

【中文标题】如何在 C# 中使用 SpInprocRecoContext 识别语音事件？

【英文标题】：How to recognize voice events using SpInprocRecoContext in C#?

【发布时间】：2018-04-13 20:37:12

【问题描述】：

我非常接近完成一个通过 C# (SAPI 5.4) 修改 Windows 语音词典的个人项目。我正在研究的最后一点是如何为给定的单词设置 SAPI 电话。我找到了一种通过 C# 表单和通过SpSharedRecoContext 获得的语音识别来执行此操作的方法。但是，我正在尝试使用语音文件（* .wav）作为输入进行识别。我知道这需要通过SpInprocRecoContext 来完成。

我从 Microsoft 中发现的关于 SAPI 5.4 识别的每个示例（例如 VB 的 this one）都是针对 SpSharedRecoContext 而不是 SpInprocRecoContext（我相信我已经看到 cmets 发现其中一些示例缺少细节）。此外，我在这个论坛上发现了多个主题（主要由 Eric Brown 回答，参见 topic 1、topic 2、topic 3）提到使用 SpInprocRecoContext 需要比 SpSharedRecoContext 更多的设置，但 我还没有找到在 C# 中使用 SpInprocRecoContext 时如何捕获语音识别事件的明确答案。

我该如何继续？

这是我目前的代码（为更好的组织而编辑）：

using SpeechLib;
using System;

namespace SpeechTest

    class Program
           
        static void Main(string[] args)
        
            string MyText = "dolphins"; // Text string of interest
            string WaveFile = @"C:\Reco\MYWAVE.wav"; // Path to wav file used for voice recognition

            // Declare important recognition objects
            SpInprocRecognizer Recognizer;
            SpInProcRecoContext RecoContext;
            ISpeechRecoGrammar grammar;
            ISpeechFileStream MyFileStream;
            ISpeechVoice MyVoice;

            // Create recognizer and recognition context
            RecoContext = new SpInProcRecoContext();
            Recognizer = (SpInprocRecognizer)RecoContext.Recognizer;            

            // Set up recognition event handling            
            RecoContext.Recognition += new _ISpeechRecoContextEvents_RecognitionEventHandler(RecoContext_Recognition);

            // Set up the grammar
            grammar = RecoContext.CreateGrammar(); // Initialize the grammar
            grammar.DictationLoad("", SpeechLoadOption.SLOStatic); // Set up dictation grammar
            grammar.DictationSetState(SpeechRuleState.SGDSActive); // Activate the grammar

            // Set up audio input for SpInProcRecoContext
            SpObjectTokenCategory Category = new SpObjectTokenCategory();
            Category.SetId(SpeechStringConstants.SpeechCategoryAudioIn);
            SpObjectToken AudioToken = new SpObjectToken();
            AudioToken.SetId(Category.Default);
            Recognizer.AudioInput = AudioToken;

            //Category.SetId(SpeechStringConstants.SpeechCategoryRecognizers); // <== generates a runtime error!!!
            //SpObjectToken EngineToken = new SpObjectToken();
            //EngineToken.SetId(Category.Default);
            //Recognizer.Recognizer = EngineToken;

            //Category.SetId(SpeechStringConstants.SpeechCategoryRecoProfiles); // <== generates a runtime error!!!
            //SpObjectToken ProfileToken = new SpObjectToken();
            //ProfileToken.SetId(Category.Default);
            //Recognizer.Profile = ProfileToken;

            // Create an audio file stream from MyText
            MyFileStream = new SpFileStream(); // Create new SpFileStream instance
            TextToWave(MyText, WaveFile); // Call function to create a wav file that voices MyText
            MyFileStream.Open(WaveFile, SpeechStreamFileMode.SSFMOpenForRead, true);

            // Activate the recognizer and input the audio file stream into the recognizer
            RecoContext.State = SpeechRecoContextState.SRCS_Enabled;
            Recognizer.State = SpeechRecognizerState.SRSActive;
            Recognizer.AudioInputStream = MyFileStream; // <== generates a runtime error!!!

            // Output info and cleanup
            Console.WriteLine(MyText + " = " + SAPIPhonemes);
            MyFileStream.Close();
            Console.ReadLine();
        

        static void TextToWave(string text, string file)
        
            SpFileStream fileStream = new SpFileStream();
            SpVoice voice = new SpVoice();

            fileStream.Open(file, SpeechStreamFileMode.SSFMCreateForWrite, true);
            voice.AudioOutputStream = fileStream;
            voice.Speak(text);

            fileStream.Close();
        

        public static string SAPIPhonemes = null;

        public static void RecoContext_Recognition(int StreamNumber, object StreamPosition, SpeechRecognitionType RecognitionType, ISpeechRecoResult Result)
        
            // This event is recognized and all the below code works fine when using SpSharedRecoContext

            Console.WriteLine(Result.ToString());
            string SAPIPhonemes = null;
            SpPhoneConverter MyPhoneConverter = new SpPhoneConverter();
            MyPhoneConverter.LanguageId = 1033;

            foreach (ISpeechPhraseElement MyPhrase in Result.PhraseInfo.Elements)
            
                SAPIPhonemes += " " + MyPhoneConverter.IdToPhone(MyPhrase.Pronunciation);
            
        
    

这里可以参考基于表单的 SpSharedRecoContext 代码：

using SpeechLib;
using System;
using System.Windows.Forms;

namespace RecoForm

    public partial class Form1 : Form
    
        // Speech Recognition Object
        SpSharedRecoContext listener;

        // Grammar object
        ISpeechRecoGrammar grammar;

        public Form1()
        
            InitializeComponent();
        

        private void Form1_Load(object sender, EventArgs e)
        
            // nothing
        

        public string ps;
        private void button1_Click(object sender, EventArgs e)
        
            if (btnListen.Text == "Start Listening")
            
               // textBox1.Clear();
                try
                

                    listener = new SpSharedRecoContext();
                    listener.Recognition += new _ISpeechRecoContextEvents_RecognitionEventHandler(listener_Reco);
                    grammar = listener.CreateGrammar(0);
                    grammar.DictationLoad("", SpeechLoadOption.SLOStatic);
                    grammar.DictationSetState(SpeechRuleState.SGDSActive);
                    btnListen.Text = "Stop Listening";
                    if (ps == "1")
                    
                        listener.Resume();
                        ps = "0";
                    
                
                catch (Exception ex)
                
                    MessageBox.Show(ex.Message);
                
            
            else if (btnListen.Text == "Stop Listening")
            
                listener.Pause();
                btnListen.Text = "Start Listening";
                if (ps == "0")
                
                    ps = "1";
                
            
                

        public void listener_Reco(int StreamNumber, object StreamPosition, SpeechRecognitionType RecognitionType, ISpeechRecoResult Result)
        
            string heard = Result.PhraseInfo.GetText(0, -1, true);
            textBox1.Text += " " + heard;

            SpPhoneConverter MyPhoneConverter = new SpPhoneConverter();
            MyPhoneConverter.LanguageId = 1033;

            foreach (ISpeechPhraseElement MyPhrase in Result.PhraseInfo.Elements)
                textBox2.Text += " " + MyPhoneConverter.IdToPhone(MyPhrase.Pronunciation);
        
    


// https://***.com/questions/11935533/c-sharp-sapi-5-4-languages

这是另一个示例（在 VB 中），它结合了仍然不起作用的 Microsoft 示例（here 和 here）（请参阅 Command1_Click 中的 cmets 以查找我遇到运行时错误的位置）。

Imports SpeechLib

Public Class Form1
    Const WaveFile = "C:\Reco\MYWAVE.wav"

    Dim WithEvents RC As SpInProcRecoContext
    Dim Recognizer As SpInprocRecognizer
    Dim myGrammar As ISpeechRecoGrammar
    Dim MyFileStream As SpeechLib.SpFileStream
    Dim MyVoice As SpeechLib.SpVoice
    Dim MyText As String

    Private Sub Form1_Load(sender As Object, e As EventArgs) Handles MyBase.Load
        On Error GoTo EH

        RC = New SpInProcRecoContext
        Recognizer = RC.Recognizer

        myGrammar = RC.CreateGrammar
        myGrammar.DictationSetState(SpeechRuleState.SGDSActive)
        MyVoice = New SpVoice
        MyVoice.Voice = MyVoice.GetVoices("gender=male").Item(0)

        Dim Category As SpObjectTokenCategory
        Category = New SpObjectTokenCategory
        Category.SetId(SpeechStringConstants.SpeechCategoryAudioIn)

        Dim Token As SpObjectToken
        Token = New SpObjectToken
        Token.SetId(Category.Default)
        Recognizer.AudioInput = Token

        TextBox1.Text = "play the eight of clubs"

EH:
        If Err.Number Then ShowErrMsg()
    End Sub

    Private Sub Command1_Click(sender As Object, e As EventArgs) Handles Command1.Click
        MyFileStream = MakeWAVFileFromText(TextBox1.Text, WaveFile)
        MyFileStream.Open(WaveFile)
        Recognizer.AudioInputStream = MyFileStream ' ==> produces a runtime error!!!
    End Sub

    Private Sub RC_Recognition(ByVal StreamNumber As Long, ByVal StreamPosition As Object, ByVal RecognitionType As SpeechLib.SpeechRecognitionType, ByVal Result As SpeechLib.ISpeechRecoResult)
        On Error GoTo EH

        TextBox2.Text = Result.PhraseInfo.GetText

EH:
        If Err.Number Then ShowErrMsg()
    End Sub

    Private Sub ShowErrMsg()

        ' Declare identifiers:
        Const NL = vbNewLine
        Dim T As String

        T = "Desc: " & Err.Description & NL
        T = T & "Err #: " & Err.Number
        MsgBox(T, vbExclamation, "Run-Time Error")
        End

    End Sub

    Private Function MakeWAVFileFromText(ByVal strText As String, ByVal strFName As String) As SpFileStream

        On Error GoTo EH

        ' Declare identifiers:
        Dim FileStream As SpFileStream
        Dim Voice As SpVoice

        ' Instantiate Voice and FileStream objects:
        Voice = New SpVoice
        FileStream = New SpFileStream

        ' Open specified .wav file, set voice output
        ' to file, and speak synchronously:
        FileStream.Open(strFName, SpeechStreamFileMode.SSFMCreateForWrite, True)
        Voice.AudioOutputStream = FileStream
        Voice.Speak(strText, SpeechVoiceSpeakFlags.SVSFIsXML)

        ' Close file and return reference to FileStream object:
        FileStream.Close()
        MakeWAVFileFromText = FileStream

EH:
        If Err.Number Then ShowErrMsg()
    End Function
End Class

' https://msdn.microsoft.com/en-us/library/ee125184%28v=vs.85%29.aspx
' https://msdn.microsoft.com/en-us/library/ee125344(v=vs.85).aspx

---

更新：所以这有效，但是流事件的结束不会触发，从而阻止 Application.Run 返回。作为一种解决方法，我可以使用一些 StopWatch 抖动来关闭所有内容，但显然这并不理想。请记住，我对 C# 还是很陌生，所以我的 cmets 可能不是 100% 准确的。

任何想法如何让结束流事件触发？

using SpeechLib;
using System;
using System.Windows.Forms;

namespace SAPITextFromVoice

    class Program
    
        // Initialize variables needed throughout this code
        static ISpeechRecoGrammar grammar; // Declare the grammar
        static SpFileStream FileStream; // Declare the voice recognition input file stream
        static string AudioPath = null; // Declare directory path to wav file
        static string GrammarPath = null; // Declare directory path to grammar file

        static void Main(string[] args)
        
            // Initialize string variable for storing the text of interest
            string MyText = "the rain in spain";

            // Store path to speech grammar XML file
            //GrammarPath = @"C:\Reco\MyGrammar.xml";

            // Store path to voice recognition input wav file
            AudioPath = @"C:\Reco\MyAudio.wav";

            TextToWav(AudioPath, MyText);

            try // Attempt the following code
            
                // Open the created wav in a new FileStream
                FileStream = new SpFileStream(); // Create new instance of SpFileStream
                FileStream.Open(AudioPath, SpeechStreamFileMode.SSFMOpenForRead, true); // Open the specified file in the FileStream for reading with events enabled

                // Create speech recognizer and associated context
                SpInprocRecognizer MyRecognizer = new SpInprocRecognizer(); // Create new instance of SpInprocRecognizer
                SpInProcRecoContext RecoContext = (SpInProcRecoContext)MyRecognizer.CreateRecoContext(); // Initialize the SpInProcRecoContext (in-process recognition context)

                // Set the voice recognition input as the FileStream
                MyRecognizer.AudioInputStream = FileStream; // This will internally "speak" the wav file for input into the voice recognition engine

                // Set up recognition event handling
                RecoContext.Recognition += new _ISpeechRecoContextEvents_RecognitionEventHandler(RecoContext_Recognition); // Register for successful voice recognition events
                RecoContext.FalseRecognition += new _ISpeechRecoContextEvents_FalseRecognitionEventHandler(RecoContext_FalseRecognition); // Register for failed (low confidence) voice recognition events
                RecoContext.Hypothesis += new _ISpeechRecoContextEvents_HypothesisEventHandler(RecoContext_Hypothesis); // Register for voice recognition hypothesis events
                RecoContext.EndStream += new _ISpeechRecoContextEvents_EndStreamEventHandler(RecoContext_EndStream); // Register for end of file stream events

                // Set up the grammar
                grammar = RecoContext.CreateGrammar(); // Initialize the grammar object
                //grammar.CmdLoadFromFile(GrammarPath, SpeechLoadOption.SLODynamic); // Load custom XML grammar file
                //grammar.CmdSetRuleIdState(0, SpeechRuleState.SGDSActive); // Activate the loaded grammar
                grammar.DictationLoad("", SpeechLoadOption.SLOStatic); // Load blank dictation topic into the grammar
                grammar.DictationSetState(SpeechRuleState.SGDSActive); // Activate dictation grammar
            
            catch // Handle exceptions in above code
            
                Console.WriteLine("Error during voice recognition setup");
                return; // Stop executing the code
            

            Application.Run(); // Starts a standard application message loop on the current thread

            Console.WriteLine("done");
            Console.ReadLine();
        

        // Function for converting text to a voiced wav file via text-to-speech
        public static bool TextToWav(string FilePath, string text)
        
            try // Attempt the following code
            
                if (System.IO.File.Exists(FilePath) == true) // Check if voice recognition wav file already exists
                    System.IO.File.Delete(FilePath); // Delete existing voice recognitoin wav file
                SpFileStream stream = new SpFileStream(); // Create new SpFileStream instance
                stream.Format.Type = SpeechAudioFormatType.SAFT48kHz16BitStereo; // Set the file stream audio format
                stream.Open(FilePath, SpeechStreamFileMode.SSFMCreateForWrite, true); // Open the specified file for writing with events enabled

                SpVoice voice = new SpVoice(); // Create new SPVoice instance
                voice.Volume = 100; // Set the volume level of the text-to-speech voice
                voice.Rate = -2; // Set the rate at which text is spoken by the text-to-speech engine
                string NameAttribute = "Name = " + "Microsoft Anna";
                voice.Voice = voice.GetVoices(NameAttribute).Item(0);
                //voice.Speak(text);
                voice.AudioOutputStream = stream; // Send the audio output to the file stream
                voice.Speak(text, SpeechVoiceSpeakFlags.SVSFDefault); // Internally "speak" the inputted text (which records it in the wav file)

                stream.Close(); // Close the file stream
                return true; // Send "true" back to calling code line
            
            catch // Handle exceptions in above code
            
                Console.WriteLine("Error during wav file creation");
                return false; // Send "false" back to calling code line
            
        

        // Event handler for successful (higher confidence) voice recognition
        public static void RecoContext_Recognition(int StreamNumber, object StreamPosition, SpeechRecognitionType RecognitionType, ISpeechRecoResult Result)
        
            RecognitionProcessing(Result, true); // Process the voice recognition result
        

        // Event handler for false (low confidence) voice recognition
        public static void RecoContext_FalseRecognition(int StreamNumber, object StreamPosition, ISpeechRecoResult Result)
        
            RecognitionProcessing(Result, false); // Process the voice recognition result
        

        // Event handler for voice recognition hypotheses
        public static void RecoContext_Hypothesis(int StreamNumber, object StreamPosition, ISpeechRecoResult Result)
        
            float confidence = Result.PhraseInfo.Elements.Item(0).EngineConfidence;
            Console.WriteLine(("Hypothesis = " + Result.PhraseInfo.GetText() + " (" + Decimal.Round(Convert.ToDecimal(confidence), (confidence > 0.01 ? 3 : 4)) + ")")); // Output info to console
        

        // Event handler for reaching the end of an audio input stream
        public static void RecoContext_EndStream(int StreamNumber, object StreamPosition, bool StreamReleased)
        
            // Clean up now that voice recognition is complete

            Console.WriteLine("--- END OF STREAM ---"); // Output info to the console

            try // Attempt the following code
            
                //grammar.CmdSetRuleIdState(0, SpeechRuleState.SGDSInactive); // Deactivate the loaded grammar
                grammar.DictationSetState(SpeechRuleState.SGDSInactive); // Deactivate dictation grammar
                FileStream.Close(); // Close the input FileStream

                Application.ExitThread(); // Terminates the message loop on the current thread
            
            catch // Handle exceptions in above code
            
                Console.WriteLine("Error during cleanup process");
            
        

        // Function for processing voice recognition results
        public static void RecognitionProcessing(ISpeechRecoResult Result, bool RecoType)
        
            try // Attempt the following code
            
                string RecognizedText = Result.PhraseInfo.GetText().Trim(); // Store recognized text    
                float confidence = Result.PhraseInfo.Elements.Item(0).EngineConfidence; // Get confidence of voice recognition result
                decimal RecognitionConfidence = Decimal.Round(Convert.ToDecimal(confidence), (confidence > 0.01 ? 3 : 4)); // Calculate confidence of voice recognition result convert to decimal, and round the result
                Console.WriteLine((RecoType == false ? "false " : "") + "recognition = " + RecognizedText + " (" + RecognitionConfidence + ")"); // Output info to the console
                GetPhonemes(Result); // Retrieve SAPI phonemes from recognized words
            
            catch // Handle exceptions in above code
            
                Console.WriteLine("Error during processing of recognition result");
            
        

        // Function for extracting SAPI phonemes from voice recognition results
        public static void GetPhonemes(ISpeechRecoResult Result)
        
            try // Attempt the following code
            
                SpPhoneConverter MyPhoneConverter = new SpPhoneConverter(); // Create new SPPhoneConverter instance
                MyPhoneConverter.LanguageId = 1033; // Set the phone converter's language (English = 1033)
                string SAPIPhonemesRaw = null; // Initialize string for storing raw SAPI phoneme data
                string SAPIPhonemes = null; // Initialize string for storing delimited SAPI phoneme data
                int i = 1; // Initialize integer for tracking phoneme count

                foreach (ISpeechPhraseElement MyPhrase in Result.PhraseInfo.Elements) // Loop through each element of the recognized text
                
                    SAPIPhonemesRaw += " " + MyPhoneConverter.IdToPhone(MyPhrase.Pronunciation); // Build string of SAPI phonemes extracted from the recognized text
                    SAPIPhonemes += (i++ > 1 ? " - " : " ") + MyPhoneConverter.IdToPhone(MyPhrase.Pronunciation); // Build string of SAPI phonemes extracted from the recognized text, delimited by "-"
                

                Console.WriteLine("Phonemes = " + SAPIPhonemes.Trim());
            
            catch // Handle exceptions in above code
            
                Console.WriteLine("Error during phoneme extraction");
            
        
    

【问题讨论】：

感谢@halfer 的编辑，感谢您关于避免无关内容的建议。

不用担心 Exergist，不客气。如果您有兴趣，可以在 Meta 上进行一些参考讨论，例如here 和 here。总之，看起来是个不错的Q！

另外，你遇到了什么运行时错误？

不是很明显这是如何发生的，System.Speech 命名空间中 SAPI 的 .NET 包装器非常好。上手容易多了。并获得帮助。

基本上我有一个语音识别应用程序，可以将 C# 代码作为宏的一部分运行。我创建了一组可以修改 Windows 语音识别字典（应用程序可以利用它来改进识别和文本到语音）的 C# 函数集合，据我所知，唯一的方法是直接通过 SAPI .我项目的最后一部分涉及识别口语短语并提取 SAPI 音素，然后将其用作字典的输入。我知道我可以使用 System.Speech 获得 IPA 发音，但不确定 SAPI。简而言之就是这个故事。

【参考方案1】：

抱歉花了这么长时间，但查看您的代码我发现了几个可能的问题。

在激活识别器之前，您需要在识别器上设置输入流。一旦识别器激活，它将立即开始读取。更改活动识别器上的输入流将导致错误。 在将识别器设置为活动之前，您确实需要设置一个记录配置文件和记录引擎。我将为每种类型创建单独的 SpObjectTokenCategory 对象。

【讨论】：

感谢 Eric 的评论，我希望您能找到我的帖子 :) 我编辑的代码与您的上述答案相比如何？

再次感谢您的帮助埃里克！您似乎对语音识别和 SAPI 了解很多，我希望您可以快速查看我的相关问题 here。这是我项目的最后一部分，非常感谢您的反馈！

【参考方案2】：

我正在回过头来提供完整的解决方案，让我可以使用给定的单词，创建带有文本到语音的浊音文件流，然后提取该单词的 SAPI 音素。包含在我原来的问题的答案。此外，using SpeechLib 指的是 Interop.SpeechLib.dll，它是 (COM) Microsoft Speech Object Library v5.4。

请记住，此代码在另一个名为 VoiceAttack 的父应用程序中用作“内联函数”，因此代码的格式与您在 Visual Studio 中的预期略有不同。从这种格式转换到 Visual Studio 并不困难，希望其他人可以将此作为未来工作的跳板。

请注意，我是 C# 爱好者。该代码在功能和速度方面完全按照我的需要工作，但它可能不像某些人喜欢的那样“优化”，并且描述性 cmets 仅限于我可用的知识。我绝对愿意接受有关如何改进它的建议。

非常感谢 Eric Brown 的反馈！

using SpeechLib;
using System;
using System.IO;
using System.Threading;
using System.Windows.Forms;

class VAInline

    // Initialize variables needed throughout this code
    ISpeechRecoGrammar grammar; // Declare the grammar
    SpFileStream FileStream; // Declare the voice recognition input file stream
    string AudioPath = null; // Declare directory path to wav file
    string GrammarPath = null; // Declare directory path to grammar file
    string RecognitionFlag = "";
    string RecognitionConfidence = "";
    bool UseDictation; // Declare boolean variable for storing pronunciation dictation grammar setting

    public void main()
    
        // Reset relevant VoiceAttack text variables
        VA.SetText("~~RecognitionError", null);
        VA.SetText("~~RecognizedText", null);
        VA.SetText("~~SAPIPhonemes", null);
        VA.SetText("~~SAPIPhonemesRaw", null);
        //VA.SetText("~~FalseRecognitionFlag", null);

        // Retrieve the desired word data contained within VoiceAttack text variable
        string ProcessText = null; // Initialize string variable for storing the text of interest
        if (VA.GetText("~~ProcessText") != null) // Check if user provided valid text in input variable
            ProcessText = VA.GetText("~~ProcessText"); // Store text of interest held by VA text variable
        else
        
            VA.SetText("~~RecognitionError", "Error in input text string (SAPI)"); // Send error detail back to VoiceAttack as text variable
            return; // End code processing
        

        // Retrieve path to speech grammar XML file from VoiceAttack
        GrammarPath = VA.GetText("~~GrammarFilePath");

        // Retrieve path to voice recognition input wav file from VoiceAttack
        AudioPath = VA.GetText("~~AudioFilePath");

        // Check if TTS engine is voicing the input for the speech recognition engine
        if (VA.GetBoolean("~~UserVoiceInput") == false)
        
            //VA.WriteToLog("creating wav file");
            if (TextToWav(AudioPath, ProcessText) == false) // Create wav file with specified path that voices specified text (with text-to-speech) and check if the creation was NOT successful
                return; // Stop executing the code
        

        // Create speech recognizer and associated context
        SpInprocRecognizer MyRecognizer = new SpInprocRecognizer(); // Create new instance of SpInprocRecognizer
        SpInProcRecoContext RecoContext = (SpInProcRecoContext)MyRecognizer.CreateRecoContext(); // Initialize the SpInProcRecoContext (in-process recognition context)

        try // Attempt the following code
        
            // Open the created wav in a new FileStream
            FileStream = new SpFileStream(); // Create new instance of SpFileStream
            FileStream.Open(AudioPath, SpeechStreamFileMode.SSFMOpenForRead, true); // Open the specified file in the FileStream for reading with events enabled

            // Set the voice recognition input as the FileStream
            MyRecognizer.AudioInputStream = FileStream; // This will internally "speak" the wav file for input into the voice recognition engine

            // Set up recognition event handling
            RecoContext.Recognition += new _ISpeechRecoContextEvents_RecognitionEventHandler(RecoContext_Recognition); // Register for successful voice recognition events
            RecoContext.FalseRecognition += new _ISpeechRecoContextEvents_FalseRecognitionEventHandler(RecoContext_FalseRecognition); // Register for failed (low confidence) voice recognition events
            if (VA.GetBoolean("~~ShowRecognitionHypothesis") == true) // Check if user wants to show voice recognition hypothesis results
                RecoContext.Hypothesis += new _ISpeechRecoContextEvents_HypothesisEventHandler(RecoContext_Hypothesis); // Register for voice recognition hypothesis events
            RecoContext.EndStream += new _ISpeechRecoContextEvents_EndStreamEventHandler(RecoContext_EndStream); // Register for end of file stream events

            // Set up the grammar
            grammar = RecoContext.CreateGrammar(); // Initialize the grammar object
            UseDictation = (bool?)VA.GetBoolean("~~UseDictation") ?? false; // Set UserDictation based on value from VoiceAttack boolean variable
            if (UseDictation == true) // Check if pronunciation dictation grammar should be used with speech recognition
            
                //grammar.DictationLoad("", SpeechLoadOption.SLOStatic); // Load blank dictation topic into the grammar
                grammar.DictationLoad("Pronunciation", SpeechLoadOption.SLOStatic); // Load pronunciation dictation topic into the grammar so that the raw (unfiltered) phonemes may be retrieved
                grammar.DictationSetState(SpeechRuleState.SGDSActive); // Activate dictation grammar
            
            else
            
                grammar.CmdLoadFromFile(GrammarPath, SpeechLoadOption.SLODynamic); // Load custom XML grammar file
                grammar.CmdSetRuleIdState(0, SpeechRuleState.SGDSActive); // Activate the loaded grammar
            
            Application.Run(); // Starts a standard application message loop on the current thread
        
        catch // Handle exceptions in above code
        
            VA.SetText("~~RecognitionError", "Error during voice recognition setup (SAPI)"); // Send error detail back to VoiceAttack as text variable
            return; // Stop executing the code
        
        finally // Runs whether an exception is encountered or not
        
            MyRecognizer = null; // Set to null in preparation for garbage collection
            FileStream.Close(); // Close the input FileStream
            FileStream = null; // Set to null in preparation for garbage collection

            // Close up recognition event handling
            RecoContext.Recognition -= new _ISpeechRecoContextEvents_RecognitionEventHandler(RecoContext_Recognition); // Unregister for successful voice recognition events
            RecoContext.FalseRecognition -= new _ISpeechRecoContextEvents_FalseRecognitionEventHandler(RecoContext_FalseRecognition); // Unregister for failed (low confidence) voice recognition events
            if (VA.GetBoolean("~~ShowRecognitionHypothesis") == true) // Check if user wanted to show voice recognition hypothesis results
                RecoContext.Hypothesis -= new _ISpeechRecoContextEvents_HypothesisEventHandler(RecoContext_Hypothesis); // Unregister for voice recognition hypothesis events
            RecoContext.EndStream -= new _ISpeechRecoContextEvents_EndStreamEventHandler(RecoContext_EndStream); // Unregister for end of file stream events
            RecoContext = null; // Set to null in preparation for garbage collection
        
        //VA.WriteToLog("voice recognition complete"); // Output info to event log
    

    // Function for converting text to a voiced wav file via text-to-speech
    public bool TextToWav(string FilePath, string text)
    
        //VA.WriteToLog("creating wav file"); // Output info to event log
        SpFileStream stream = new SpFileStream(); // Create new SpFileStream instance
        try // Attempt the following code
        
            if (System.IO.File.Exists(FilePath) == true) // Check if voice recognition wav file already exists
                System.IO.File.Delete(FilePath); // Delete existing voice recognition wav file
            stream.Format.Type = SpeechAudioFormatType.SAFT48kHz16BitStereo; // Set the file stream audio format
            stream.Open(FilePath, SpeechStreamFileMode.SSFMCreateForWrite, true); // Open the specified file for writing with events enabled
            SpVoice voice = new SpVoice(); // Create new SPVoice instance
            voice.Volume = 100; // Set the volume level of the text-to-speech voice
            voice.Rate = -2; // Set the rate at which text is spoken by the text-to-speech engine
            string NameAttribute = "Name = " + VA.GetText("~~TextToSpeechVoice");
            voice.Voice = voice.GetVoices(NameAttribute).Item(0);
            //voice.Speak(text);
            voice.AudioOutputStream = stream; // Send the audio output to the file stream
            voice.Speak(text, SpeechVoiceSpeakFlags.SVSFDefault); // Internally "speak" the inputted text (which records it in the wav file)
            voice = null; // Set to null in preparation for garbage collection
        
        catch // Handle exceptions in above code
        
            VA.SetText("~~RecognitionError", "Error during wav file creation (SAPI)"); // Send error detail back to VoiceAttack as text variable
            return false; // Send "false" back to calling code line
        
        finally // Runs whether an exception is encountered or not
        
            stream.Close(); // Close the file stream
            stream = null; // Set to null in preparation for garbage collection
        
        return true; // Send "true" back to calling code line
    

    // Event handler for successful (higher confidence) voice recognition
    public void RecoContext_Recognition(int StreamNumber, object StreamPosition, SpeechRecognitionType RecognitionType, ISpeechRecoResult Result)
    
        //VA.WriteToLog("Recognition successful"); // Output info to event log

        //VA.SetText("~~FalseRecognitionFlag", ""); // Send blank recognition flag ("") back to VoiceAttack as text variable
        //RecognitionFlag = ""; // Set the RecognitionFlag as blank
        RecognitionProcessing(Result); // Process the voice recognition result
        //if (UseDictation == false) // Check if pronunciation dictation grammar should NOT be used with speech recognition
        GetPhonemes(Result); // Retrieve SAPI phonemes from recognition result
    

    // Event handler for unsuccessful (low confidence) voice recognition
    public void RecoContext_FalseRecognition(int StreamNumber, object StreamPosition, ISpeechRecoResult Result)
    
        //VA.WriteToLog("Low confidence recognition"); // Output info to event log

        //VA.WriteToLog(Result.PhraseInfo.GetText());
        //VA.SetText("~~FalseRecognitionFlag", "*"); // Send unsuccessful recognition flag (text character) back to VoiceAttack as text variable
        RecognitionFlag = "*"; // Set the RecognitionFlag as "*"
        RecognitionProcessing(Result); // Process the voice recognition result
        GetPhonemes(Result); // Retrieve SAPI phonemes from recognition result
    

    // Event handler for voice recognition hypotheses
    public void RecoContext_Hypothesis(int StreamNumber, object StreamPosition, ISpeechRecoResult Result)
    
        //VA.WriteToLog("Recognition hypothesis"); // Output info to event log

        float confidence = Result.PhraseInfo.Elements.Item(0).EngineConfidence;
        VA.WriteToLog("Hypothesis = " + Result.PhraseInfo.GetText() + " (" + Decimal.Round(Convert.ToDecimal(confidence), (confidence > 0.01 ? 3 : 4)) + ")"); // Output info to event log
    

    // Event handler for reaching the end of an audio input stream
    public void RecoContext_EndStream(int StreamNumber, object StreamPosition, bool StreamReleased)
    
        // VA.WriteToLog("End of stream, cleaning up now"); // Output info to event log

        // Clean up now that voice recognition is complete
        try // Attempt the following code
        
            if (UseDictation == true)
                grammar.DictationSetState(SpeechRuleState.SGDSInactive); // Deactivate dictation grammar
            else
                grammar.CmdSetRuleIdState(0, SpeechRuleState.SGDSInactive); // Deactivate the loaded grammar
        
        catch // Handle exceptions in above code
        
            VA.SetText("~~RecognitionError", "Error during cleanup process (SAPI)"); // Send error detail back to VoiceAttack as text variable
        
        finally // Runs whether an exception is encountered or not
        
            Application.ExitThread(); // Terminates the message loop on the current thread
        
    

    // Function for processing voice recognition results
    public void RecognitionProcessing(ISpeechRecoResult Result)
    
        //VA.WriteToLog("Processing recognition result"); // Output info to event log

        try // Attempt the following code
        
            string RecognizedText = Result.PhraseInfo.GetText().Trim(); // Store recognized text    
            float confidence = Result.PhraseInfo.Elements.Item(0).EngineConfidence; // Get confidence of voice recognition result
            decimal RecognitionConfidenceScore = Decimal.Round(Convert.ToDecimal(confidence), (confidence > 0.01 ? 3 : 4)); // Calculate confidence of voice recognition result convert to decimal, and round the result
            string RecognitionConfidenceLevel = Result.PhraseInfo.Elements.Item(0).ActualConfidence.ToString().Replace("SEC", "").Replace("Confidence", "");
            VA.SetText("~~RecognizedText", RecognizedText); // Send recognized text back to VoiceAttack as text variable
            //VA.SetText("~~RecognitionConfidenceLevel", RecognitionConfidenceLevel); // Send speech recognition confidence level back to VoiceAttack as text variable
            //VA.SetDecimal("~~RecognitionConfidence", RecognitionConfidenceScore); // Send recognized confidence back to VoiceAttack as decimal variable

            if (VA.GetBoolean("~~ShowConfidence") == true)
                RecognitionConfidence = "(" + RecognitionConfidenceLevel + " @ " + RecognitionConfidenceScore.ToString() + ")" + RecognitionFlag;
            //VA.SetText("~~RecognitionConfidence", RecognitionConfidenceLevel + " @ " + RecognitionConfidenceScore.ToString()); // Send speech recognition confidence data back to VoiceAttack as text variable
            VA.SetText("~~RecognitionConfidence", RecognitionConfidence); // Send formatted speech recognition confidence data back to VoiceAttack as text variable
            if (UseDictation == true) // Check if pronunciation dictation grammar should be used with speech recognition
            
                RecognizedText = RecognizedText.Replace("hh", "h"); // Replace any instances of "hh" in recognized phonemes with "h"
                VA.SetText("~~SAPIPhonemes", RecognizedText); // Send word-delimited SAPI phoneme data back to VoiceAttack as text variable
            
        
        catch (Exception e) // Handle exceptions in above code
        
            VA.WriteToLog(e.ToString());
            VA.SetText("~~RecognitionError", "Error during processing of recognition result (SAPI)"); // Send error detail back to VoiceAttack as text variable
        
    

    // Function for extracting SAPI phonemes from voice recognition results
    public void GetPhonemes(ISpeechRecoResult Result)
    
        //VA.WriteToLog("Extracting phonemes from voice recognition result"); // Output info to event log

        try // Attempt the following code
        
            SpPhoneConverter MyPhoneConverter = new SpPhoneConverter(); // Create new SPPhoneConverter instance
            MyPhoneConverter.LanguageId = 1033; // Set the phone converter's language (English = 1033)
            string SAPIPhonemesRaw = null; // Initialize string for storing raw SAPI phoneme data
            string SAPIPhonemes = null; // Initialize string for storing delimited SAPI phoneme data
            int i = 1; // Initialize integer for tracking phoneme count
            string WordSeparator = " "; // Initialize string variable for storing the characters used to separate words within the phoneme result

            if (VA.GetBoolean("~~SeparatePhonemes") == true) // Check if user wants to have the "-" character separate the words within the phoneme result
                WordSeparator = " - "; // Redefine the WordSeparator            
            foreach (ISpeechPhraseElement MyPhrase in Result.PhraseInfo.Elements) // Loop through each element of the recognized text
            
                if (MyPhrase.DisplayText != " ")
                
                    SAPIPhonemesRaw += " " + MyPhoneConverter.IdToPhone(MyPhrase.Pronunciation); // Build string of SAPI phonemes extracted from the recognized text
                    SAPIPhonemes += (i++ > 1 ? WordSeparator : " ") + MyPhoneConverter.IdToPhone(MyPhrase.Pronunciation); // Build string of SAPI phonemes extracted from the recognized text, delimited by " "
                
            
            MyPhoneConverter = null; // Set to null in preparation for garbage collection

            VA.SetText("~~SAPIPhonemesRaw", SAPIPhonemesRaw.Trim()); // Send raw SAPI phoneme data back to VoiceAttack as text variable
            VA.SetText("~~SAPIPhonemes", SAPIPhonemes.Trim()); // Send word-delimited SAPI phoneme data back to VoiceAttack as text variable
        
        catch // Handle exceptions in above code
        
            VA.SetText("~~RecognitionError", "Error during phoneme extraction"); // Send error detail back to VoiceAttack as text variable
        
    


// References:
// https://github.com/rti7743/rtilabs/blob/master/files/asobiba/DictationFilter/DictationFilter/SpeechRecognitionRegexp.cs
// https://***.com/questions/6193874/help-with-sapi-v5-1-speechrecognitionengine-always-gives-same-wrong-result-with/6203533#6203533
// http://www.drdobbs.com/com-objects-c-and-the-microsoft-speech-a/184416575
// http://vbcity.com/forums/t/125150.aspx
// https://people.kth.se/~maguire/DEGREE-PROJECT-REPORTS/050702-Johan_Sverin-with-cover.pdf
// https://msdn.microsoft.com/en-us/library/ee125471(v=vs.85).aspx
// https://***.com/questions/20770593/speech-to-phoneme-in-net