Fixed the C api calls and created the TTS project file (#1324)

Co-authored-by: Michael Twohey <mtwohey@americanambulance.com>

dotnet-examples/TTS/Program.cs

@@ -14,9 +14,13 @@ internal class Program
             provider = "cpu",
             vits = new SherpaOnnxOfflineTtsVitsModelConfig
             {
-                lexicon = "vits-zh-aishell3/lexicon.txt",
+                //lexicon = "vits-zh-aishell3/lexicon.txt",
-                model = "vits-zh-aishell3/vits-aishell3.onnx",
+                //model = "vits-zh-aishell3/vits-aishell3.onnx",
-                tokens = "vits-zh-aishell3/tokens.txt",
+                //tokens = "vits-zh-aishell3/tokens.txt",
+                model = @"C:\Services\Sherpa\model.onnx",
+                lexicon = "",
+                tokens = @"C:\Services\Sherpa\tokens.txt",
+                data_dir = @"C:\Services\Sherpa\espeak-ng-data",
 
                 noise_scale = 0.667f,
                 noise_scale_w = 0.8f,

dotnet-examples/TTS/Struct/SherpaOnnxOfflineTtsConfig.cs

@@ -6,5 +6,13 @@ namespace TTS.Struct
     public struct SherpaOnnxOfflineTtsConfig
     {
         public SherpaOnnxOfflineTtsModelConfig model;
+
+        [MarshalAs(UnmanagedType.LPStr)]
+        public string rule_fsts;
+
+        public int max_num_sentences;
+
+        [MarshalAs(UnmanagedType.LPStr)]
+        public string rule_fars;
     }
 }

dotnet-examples/TTS/Struct/SherpaOnnxOfflineTtsVitsModelConfig.cs

@@ -21,6 +21,9 @@ namespace TTS.Struct
         [MarshalAs(UnmanagedType.LPStr)]
         public string tokens;
 
+        [MarshalAs(UnmanagedType.LPStr)]
+        public string data_dir;
+
         /// <summary>
         /// VITS模型的noise_scale (float，默认值= 0.667)
         /// </summary>
@@ -34,6 +37,9 @@ namespace TTS.Struct
         /// </summary>
         public float length_scale = 1f;
 
+        [MarshalAs(UnmanagedType.LPStr)]
+        public string dict_dir;
+
         public SherpaOnnxOfflineTtsVitsModelConfig()
         {
             noise_scale = 0.667f;
@@ -43,6 +49,8 @@ namespace TTS.Struct
             model = "vits-zh-aishell3/vits-aishell3.onnx";
             lexicon = "vits-zh-aishell3/lexicon.txt";
             tokens = "vits-zh-aishell3/tokens.txt";
+            data_dir = "";
+            dict_dir = "";
         }
     }
 }

dotnet-examples/TTS/TTS.csproj

@@ -0,0 +1,15 @@
+<Project Sdk="Microsoft.NET.Sdk">
+
+  <PropertyGroup>
+    <OutputType>Exe</OutputType>
+    <TargetFramework>net6.0</TargetFramework>
+    <ImplicitUsings>enable</ImplicitUsings>
+    <Nullable>enable</Nullable>
+	<AllowUnsafeBlocks>true</AllowUnsafeBlocks>
+  </PropertyGroup>
+  <ItemGroup>
+    <PackageReference Include="NAudio" Version="2.2.1" />
+    <PackageReference Include="org.k2fsa.sherpa.onnx" Version="1.10.23" />
+  </ItemGroup>
+  
+</Project>

dotnet-examples/TTS/TTSCore.cs

@@ -7,7 +7,7 @@ namespace TTS
     {
         public const string Filename = "sherpa-onnx-c-api";
 
-        [DllImport(Filename)]
+        [DllImport(Filename, CallingConvention = CallingConvention.Cdecl)]
         private static extern IntPtr SherpaOnnxCreateOfflineTts(SherpaOnnxOfflineTtsConfig handle);
 
         [DllImport(Filename)]
@@ -23,7 +23,12 @@ namespace TTS
 
         public TTSCore(SherpaOnnxOfflineTtsConfig modelConfig)
         {
-            thisHandle = SherpaOnnxCreateOfflineTts(modelConfig);
+          IntPtr ttsHandle = SherpaOnnxCreateOfflineTts(modelConfig);
+          if (ttsHandle == IntPtr.Zero)
+          {
+            throw new InvalidOperationException("Failed to create SherpaOnnx TTS engine.");
+          }
+          thisHandle = ttsHandle;
         }
 
         /// <summary>

dotnet-examples/sherpa-onnx.sln

@@ -23,11 +23,13 @@ Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "offline-punctuation", "offl
 EndProject
 Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "vad-non-streaming-asr-paraformer", "vad-non-streaming-asr-paraformer\vad-non-streaming-asr-paraformer.csproj", "{8CD6B7E5-F59F-47B3-BB87-2B2E3678924D}"
 EndProject
-Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Common", "Common\Common.csproj", "{401E963F-E25A-43CE-987D-8DB2D4715756}"
+Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "Common", "Common\Common.csproj", "{401E963F-E25A-43CE-987D-8DB2D4715756}"
 EndProject
-Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "keyword-spotting-from-files", "keyword-spotting-from-files\keyword-spotting-from-files.csproj", "{A87EDD31-D654-4C9F-AED7-F6F2825659BD}"
+Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "keyword-spotting-from-files", "keyword-spotting-from-files\keyword-spotting-from-files.csproj", "{A87EDD31-D654-4C9F-AED7-F6F2825659BD}"
 EndProject
-Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "keyword-spotting-from-microphone", "keyword-spotting-from-microphone\keyword-spotting-from-microphone.csproj", "{AEE0ED2B-C86F-4952-863C-EAD3219CB4EC}"
+Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "keyword-spotting-from-microphone", "keyword-spotting-from-microphone\keyword-spotting-from-microphone.csproj", "{AEE0ED2B-C86F-4952-863C-EAD3219CB4EC}"
+EndProject
+Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "TTS", "TTS\TTS.csproj", "{DACE4A18-4FC8-4437-92BF-5A90BA81286C}"
 EndProject
 Global
 	GlobalSection(SolutionConfigurationPlatforms) = preSolution
@@ -87,6 +89,10 @@ Global
 		{AEE0ED2B-C86F-4952-863C-EAD3219CB4EC}.Debug|Any CPU.Build.0 = Debug|Any CPU
 		{AEE0ED2B-C86F-4952-863C-EAD3219CB4EC}.Release|Any CPU.ActiveCfg = Release|Any CPU
 		{AEE0ED2B-C86F-4952-863C-EAD3219CB4EC}.Release|Any CPU.Build.0 = Release|Any CPU
+		{DACE4A18-4FC8-4437-92BF-5A90BA81286C}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+		{DACE4A18-4FC8-4437-92BF-5A90BA81286C}.Debug|Any CPU.Build.0 = Debug|Any CPU
+		{DACE4A18-4FC8-4437-92BF-5A90BA81286C}.Release|Any CPU.ActiveCfg = Release|Any CPU
+		{DACE4A18-4FC8-4437-92BF-5A90BA81286C}.Release|Any CPU.Build.0 = Release|Any CPU
 	EndGlobalSection
 	GlobalSection(SolutionProperties) = preSolution
 		HideSolutionNode = FALSE