Add endpointer mode to java and csharp API

android/build.gradle

@@ -10,7 +10,7 @@ buildscript {
 }
 
 allprojects {
-    version = '0.3.50'
+    version = '0.3.70'
 }
 
 subprojects {

android/lib/src/main/java/org/vosk/LibVosk.java

@@ -56,6 +56,10 @@ public class LibVosk {
 
     public static native void vosk_recognizer_reset(Pointer recognizer);
 
+    public static native void vosk_recognizer_set_endpointer_mode(Pointer recognizer, int mode);
+
+    public static native void vosk_recognizer_set_endpointer_delays(Pointer recognizer, float t_start_max, float t_end, float t_max);
+
     public static native void vosk_recognizer_free(Pointer recognizer);
 
     public static native Pointer vosk_text_processor_new(String verbalizer, String tagger);

android/lib/src/main/java/org/vosk/Recognizer.java

@@ -236,6 +236,34 @@ public class Recognizer extends PointerType implements AutoCloseable {
         LibVosk.vosk_recognizer_reset(this.getPointer());
     }
 
+    /**
+     * Endpointer delay mode
+     */
+    public class EndpointerMode {
+        public static final int DEFAULT = 0;
+        public static final int SHORT = 1;
+        public static final int LONG = 2;
+        public static final int VERY_LONG = 3;
+    }
+
+    /**
+     * Configures endpointer mode for recognizer
+     */
+    public void setEndpointerMode(int mode) {
+        LibVosk.vosk_recognizer_set_endpointer_mode(this.getPointer(), mode);
+    }
+
+    /**
+     * Set endpointer delays
+     *
+     * @param t_start_max     timeout for stopping recognition in case of initial silence (usually around 5.0)
+     * @param t_end           timeout for stopping recognition in milliseconds after we recognized something (usually around 0.5 - 1.0)
+     * @param t_max           timeout for forcing utterance end in milliseconds (usually around 20-30)
+     **/
+    public void setEndpointerDelays(float t_start_max, float t_end, float t_max) {
+        LibVosk.vosk_recognizer_set_endpointer_delays(this.getPointer(), t_start_max, t_end, t_max);
+    }
+
     /**
      * Releases recognizer object.
      * Underlying model is also unreferenced and if needed, released.

csharp/demo/VoskDemo.cs

@@ -28,6 +28,9 @@ public class VoskDemo
    {
         // Demo float array
         VoskRecognizer rec = new VoskRecognizer(model, 16000.0f);
+
+        rec.SetEndpointerMode(EndpointerMode.LONG);
+
         using(Stream source = File.OpenRead("test.wav")) {
             byte[] buffer = new byte[4096];
             int bytesRead;

csharp/demo/VoskDemo.csproj

@@ -2,7 +2,7 @@
 
   <PropertyGroup>
     <OutputType>Exe</OutputType>
-    <TargetFramework>net5.0</TargetFramework>
+    <TargetFramework>net8.0</TargetFramework>
     <RootNamespace>VoskDemo</RootNamespace>
   </PropertyGroup>
 
@@ -11,7 +11,7 @@
   </PropertyGroup>
 
   <ItemGroup>
-    <PackageReference Include="Vosk" Version="0.3.50" />
+    <PackageReference Include="Vosk" Version="0.3.70" />
   </ItemGroup>
 
 </Project>

csharp/nuget/Vosk.nuspec

@@ -2,7 +2,7 @@
 <package>
   <metadata>
     <id>Vosk</id>
-    <version>0.3.50</version>
+    <version>0.3.70</version>
     <authors>Alpha Cephei Inc</authors>
     <owners>Alpha Cephei Inc</owners>
     <license type="expression">Apache-2.0</license>
@@ -23,10 +23,10 @@ Vosk scales from small devices like Raspberry Pi or Android smartphone to big cl
     <copyright>Copyright 2020-2050 Alpha Cephei Inc</copyright>
     <tags>speech recognition voice stt asr speech-to-text ai offline privacy</tags>
     <dependencies>
-      <group targetFramework=".NETStandard2.0"/>
+      <group targetFramework="net8.0"/>
     </dependencies>
   </metadata>
   <files>
-    <file src="**" exclude="src/*.cs;build.sh;**/.keep-me;*.nupkg" />
+    <file src="**" exclude="bin/**;obj/**;build.sh;src/*.cs;*.nupkg;**/.keep-me" />
   </files>
 </package>

csharp/nuget/build.sh

@@ -1,2 +1,2 @@
-mcs -out:lib/netstandard2.0/Vosk.dll -target:library src/*.cs
-nuget pack
+rm -rf bin lib obj
+/home/shmyrev/local/dotnet/dotnet pack Vosk.csproj -p:NuspecFile=Vosk.nuspec -o .

csharp/nuget/src/VoskPINVOKE.cs

@@ -65,6 +65,12 @@ class VoskPINVOKE {
   [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_reset")]
   public static extern void VoskRecognizer_Reset(global::System.Runtime.InteropServices.HandleRef jarg1);
 
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_set_endpointer_mode")]
+  public static extern void VoskRecognizer_SetEndpointerMode(global::System.Runtime.InteropServices.HandleRef jarg1, int jarg2);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_set_endpointer_delays")]
+  public static extern void VoskRecognizer_SetEndpointerDelays(global::System.Runtime.InteropServices.HandleRef jarg1, float jarg2, float jarg3, float jarg4);
+
   [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_set_log_level")]
   public static extern void SetLogLevel(int jarg1);
 
@@ -107,7 +113,6 @@ class VoskPINVOKE {
   [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint = "vosk_batch_recognizer_get_pending_chunks")]
   public static extern int VoskBatchRecognizer_GetPendingChunks(global::System.Runtime.InteropServices.HandleRef jarg1);
 
-
-    }
+}
 
 }

csharp/nuget/src/VoskRecognizer.cs

@@ -1,5 +1,12 @@
 namespace Vosk {
 
+public enum EndpointerMode {
+    DEFAULT = 0,
+    SHORT = 1,
+    LONG = 2,
+    VERY_LONG = 3
+}
+
 public class VoskRecognizer : System.IDisposable {
   private System.Runtime.InteropServices.HandleRef handle;
 
@@ -91,6 +98,14 @@ public class VoskRecognizer : System.IDisposable {
     VoskPINVOKE.VoskRecognizer_Reset(handle);
   }
 
+  public void SetEndpointerMode(EndpointerMode mode) {
+    VoskPINVOKE.VoskRecognizer_SetEndpointerMode(handle, (int) mode);
+  }
+
+  public void SetEndpointerDelays(float t_start_max, float t_end, float t_max) {
+    VoskPINVOKE.VoskRecognizer_SetEndpointerDelays(handle, t_start_max, t_end, t_max);
+  }
+
 }
 
 }

java/lib/src/main/java/org/vosk/LibVosk.java

@@ -82,6 +82,10 @@ public class LibVosk {
 
     public static native void vosk_recognizer_reset(Pointer recognizer);
 
+    public static native void vosk_recognizer_set_endpointer_mode(Pointer recognizer, int mode);
+
+    public static native void vosk_recognizer_set_endpointer_delays(Pointer recognizer, float t_start_max, float t_end, float t_max);
+
     public static native void vosk_recognizer_free(Pointer recognizer);
 
     public static native Pointer vosk_text_processor_new(String verbalizer, String tagger);

java/lib/src/main/java/org/vosk/Recognizer.java

@@ -236,6 +236,34 @@ public class Recognizer extends PointerType implements AutoCloseable {
         LibVosk.vosk_recognizer_reset(this.getPointer());
     }
 
+    /**
+     * Endpointer delay mode
+     */
+    public class EndpointerMode {
+        public static final int DEFAULT = 0;
+        public static final int SHORT = 1;
+        public static final int LONG = 2;
+        public static final int VERY_LONG = 3;
+    }
+
+    /**
+     * Configures endpointer mode for recognizer
+     */
+    public void setEndpointerMode(int mode) {
+        LibVosk.vosk_recognizer_set_endpointer_mode(this.getPointer(), mode);
+    }
+
+    /**
+     * Set endpointer delays
+     *
+     * @param t_start_max     timeout for stopping recognition in case of initial silence (usually around 5.0)
+     * @param t_end           timeout for stopping recognition in milliseconds after we recognized something (usually around 0.5 - 1.0)
+     * @param t_max           timeout for forcing utterance end in milliseconds (usually around 20-30)
+     **/
+    public void setEndpointerDelays(float t_start_max, float t_end, float t_max) {
+        LibVosk.vosk_recognizer_set_endpointer_delays(this.getPointer(), t_start_max, t_end, t_max);
+    }
+
     /**
      * Releases recognizer object.
      * Underlying model is also unreferenced and if needed, released.

java/lib/src/test/java/org/vosk/test/DecoderTest.java

@@ -15,6 +15,7 @@ import javax.sound.sampled.UnsupportedAudioFileException;
 
 import org.vosk.LogLevel;
 import org.vosk.Recognizer;
+import org.vosk.Recognizer.EndpointerMode;
 import org.vosk.LibVosk;
 import org.vosk.Model;
 import org.vosk.TextProcessor;
@@ -96,6 +97,15 @@ public class DecoderTest {
         Assert.assertTrue(true);
     }
 
+    @Test
+    public void decoderEndpointerDelays() throws IOException, UnsupportedAudioFileException {
+        try (Model model = new Model("model");
+            Recognizer recognizer = new Recognizer(model, 16000)) {
+            recognizer.setEndpointerMode(EndpointerMode.VERY_LONG);
+            recognizer.setEndpointerDelays(5.0f, 3.0f, 50.0f);
+	}
+        Assert.assertTrue(true);
+    }
 
     @Test(expected = IOException.class)
     public void decoderTestException() throws IOException {

kotlin/src/nativeInterop/cinterop/libvosk.def

@@ -22,4 +22,4 @@ linkerOpts.linux = \
 
 linkerOpts.linux_x64 = \
 	-L/usr/lib64/ \
-	-L/usr/local/lib64/
+	-L/usr/local/lib64/

python/setup.py

@@ -45,7 +45,7 @@ with open("README.md", "rb") as fh:
 
 setuptools.setup(
     name="vosk",
-    version="0.3.50",
+    version="0.3.70",
     author="Alpha Cephei Inc",
     author_email="contact@alphacephei.com",
     description="Offline open source speech recognition API based on Kaldi and Vosk",