add java wrapper suppport (#117)

java-api-examples/Makefile

@@ -0,0 +1,61 @@
+
+ENTRY_POINT = ./
+
+LIB_SRC_DIR := ../sherpa-onnx/java-api/src/com/k2fsa/sherpa/onnx
+
+LIB_FILES = \
+	$(LIB_SRC_DIR)/EndpointRule.java \
+	$(LIB_SRC_DIR)/EndpointConfig.java \
+	$(LIB_SRC_DIR)/FeatureConfig.java \
+	$(LIB_SRC_DIR)/OnlineTransducerModelConfig.java \
+	$(LIB_SRC_DIR)/OnlineRecognizerConfig.java \
+	$(LIB_SRC_DIR)/OnlineStream.java \
+	$(LIB_SRC_DIR)/OnlineRecognizer.java \
+
+LIB_BUILD_DIR = ./lib
+
+
+EXAMPLE_FILE = DecodeFile.java
+
+JAVAC = javac
+
+BUILD_DIR = build
+
+
+RUNJFLAGS = -Dfile.encoding=utf-8
+	
+	
+vpath %.class $(BUILD_DIR)
+vpath %.java src
+	
+	
+buildexample:  
+		$(JAVAC) -cp lib/sherpaonnx.jar -d $(BUILD_DIR) -encoding UTF-8 src/$(EXAMPLE_FILE)
+	
+rebuild: clean all
+	
+.PHONY:  clean run
+	
+clean:
+	rm -frv  $(BUILD_DIR)/*
+	rm -frv  $(LIB_BUILD_DIR)/*
+	mkdir -p $(BUILD_DIR)
+	mkdir -p ./lib
+
+	
+run:
+
+		java -cp ./lib/sherpaonnx.jar:build $(RUNJFLAGS)  DecodeFile 
+
+
+buildlib: $(LIB_FILES:.java=.class)
+	
+
+%.class: %.java
+
+				$(JAVAC) -cp $(BUILD_DIR) -d $(BUILD_DIR) -encoding UTF-8 $<
+
+packjar:
+				jar cvfe lib/sherpaonnx.jar . -C $(BUILD_DIR) .
+				
+all: clean buildlib packjar buildexample run

java-api-examples/README.md

@@ -0,0 +1,88 @@
+
+0.Introduction
+---
+Java wrapper `com.k2fsa.sherpaonnx.OnlineRecognizer` for `sherpa-onnx`. Java is a cross-platform language; you can build jni .so lib according to your system, and then use the same java api for all your platform.
+``` xml
+Depend on:
+  Openjdk 1.8
+```
+---
+1.Compile libsherpa-onnx-jni.so
+---
+Compile sherpa-onnx/jni/jni.cc according to your system.
+Example for Ubuntu 18.04 LTS, Openjdk 1.8.0_362:
+``` xml
+  git clone https://github.com/k2-fsa/sherpa-onnx
+  cd sherpa-onnx
+  mkdir build
+  cd build
+  cmake -DCMAKE_BUILD_TYPE=Release -DBUILD_SHARED_LIBS=ON -DSHERPA_ONNX_ENABLE_JNI=ON ..
+  make -j6
+```
+---
+2.Download asr model files
+---
+[click here for more detail](https://k2-fsa.github.io/sherpa/onnx/pretrained_models/index.html)
+---
+3.Config model config.cfg
+---
+``` xml 
+  #model config  
+  sample_rate=16000                  
+  feature_dim=80
+  rule1_min_trailing_silence=2.4
+  rule2_min_trailing_silence=1.2
+  rule3_min_utterance_length=20
+  encoder=/sherpa-onnx/build/bin/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/encoder-epoch-99-avg-1.onnx
+  decoder=/sherpa-onnx/build/bin/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/decoder-epoch-99-avg-1.onnx
+  joiner=/sherpa-onnx/build/bin/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/joiner-epoch-99-avg-1.onnx
+  tokens=/sherpa-onnx/build/bin/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/tokens.txt
+  num_threads=4
+  enable_endpoint_detection=false
+  decoding_method=greedy_search
+  max_active_paths=4
+```
+---
+4.A simple java example
+---
+refer to [java_api_example](https://github.com/k2-fsa/sherpa-onnx/blob/master/java-api-examples/src/DecodeFile.java) for more detail.
+``` java
+    import com.k2fsa.sherpa.onnx.OnlineRecognizer;
+    import com.k2fsa.sherpa.onnx.OnlineStream;
+    String cfgpath=appdir+"/modelconfig.cfg";
+    OnlineRecognizer.setSoPath(soPath);   //set so lib path
+
+    OnlineRecognizer rcgOjb = new OnlineRecognizer();   //create a recognizer
+    rcgOjb = new OnlineRecognizer(cfgFile);    //set model config file 
+    CreateStream streamObj=rcgOjb.CreateStream();       //create a stream for read wav data
+    float[] buffer = rcgOjb.readWavFile(wavfilename); // read data from file
+    streamObj.acceptWaveform(buffer); // feed stream with data
+    streamObj.inputFinished(); // tell engine you done with all data
+    OnlineStream ssObj[] = new OnlineStream[1];
+    while (rcgOjb.isReady(streamObj)) { // engine is ready for unprocessed data
+                ssObj[0] = streamObj;
+                rcgOjb.decodeStreams(ssObj); // decode for multiple stream
+                // rcgOjb.DecodeStream(streamObj);   // decode for single stream
+            }
+
+    String recText = "simple:" + rcgOjb.getResult(streamObj) + "\n";
+    byte[] utf8Data = recText.getBytes(StandardCharsets.UTF_8);
+    System.out.println(new String(utf8Data));
+    rcgOjb.reSet(streamObj);
+    rcgOjb.releaseStream(streamObj); // release stream
+    rcgOjb.release(); // release recognizer
+       
+
+```
+---
+5.Makefile 
+---
+package jar and run app example
+package path: /sherpa-onnx/java-api-examples/lib/sherpaonnx.jar
+``` bash
+    cd sherpa-onnx/java-api-examples
+    make all          
+	
+  ```
+
+

java-api-examples/modelconfig.cfg

@@ -0,0 +1,14 @@
+#model config
+sample_rate=16000                  
+feature_dim=80
+rule1_min_trailing_silence=2.4
+rule2_min_trailing_silence=1.2
+rule3_min_utterance_length=20
+encoder=/sherpa-onnx/build_old/bin/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/encoder-epoch-99-avg-1.onnx
+decoder=/sherpa-onnx/build_old/bin/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/decoder-epoch-99-avg-1.onnx
+joiner=/sherpa-onnx/build_old/bin/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/joiner-epoch-99-avg-1.onnx
+tokens=/sherpa-onnx/build_old/bin/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/tokens.txt
+num_threads=4
+enable_endpoint_detection=false
+decoding_method=modified_beam_search
+max_active_paths=4

java-api-examples/src/DecodeFile.java

@@ -0,0 +1,174 @@
+/*
+ * // Copyright 2022-2023 by zhaoming
+ */
+/*
+Config modelconfig.cfg
+  sample_rate=16000
+  feature_dim=80
+  rule1_min_trailing_silence=2.4
+  rule2_min_trailing_silence=1.2
+  rule3_min_utterance_length=20
+  encoder=/sherpa-onnx/build/bin/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/encoder-epoch-99-avg-1.onnx
+  decoder=/sherpa-onnx/build/bin/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/decoder-epoch-99-avg-1.onnx
+  joiner=/sherpa-onnx/build/bin/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/joiner-epoch-99-avg-1.onnx
+  tokens=/sherpa-onnx/build/bin/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/tokens.txt
+  num_threads=4
+  enable_endpoint_detection=false
+  decoding_method=greedy_search
+  max_active_paths=4
+*/
+
+import com.k2fsa.sherpa.onnx.OnlineRecognizer;
+import com.k2fsa.sherpa.onnx.OnlineStream;
+import java.io.*;
+import java.nio.charset.StandardCharsets;
+
+public class DecodeFile {
+  OnlineRecognizer rcgOjb;
+  OnlineStream streamObj;
+  String wavfilename;
+
+  public DecodeFile(String fileName) {
+    wavfilename = fileName;
+  }
+
+  public void initModelWithPara() {
+    try {
+      String modelDir =
+          "/sherpa-onnx/build_old/bin/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20";
+      String encoder = modelDir + "/encoder-epoch-99-avg-1.onnx";
+      String decoder = modelDir + "/decoder-epoch-99-avg-1.onnx";
+      String joiner = modelDir + "/joiner-epoch-99-avg-1.onnx";
+      String tokens = modelDir + "/tokens.txt";
+      int numThreads = 4;
+      int sampleRate = 16000;
+      int featureDim = 80;
+      boolean enableEndpointDetection = false;
+      float rule1MinTrailingSilence = 2.4F;
+      float rule2MinTrailingSilence = 1.2F;
+      float rule3MinUtteranceLength = 20F;
+      String decodingMethod = "greedy_search";
+      int maxActivePaths = 4;
+
+      rcgOjb =
+          new OnlineRecognizer(
+              tokens,
+              encoder,
+              decoder,
+              joiner,
+              numThreads,
+              sampleRate,
+              featureDim,
+              enableEndpointDetection,
+              rule1MinTrailingSilence,
+              rule2MinTrailingSilence,
+              rule3MinUtteranceLength,
+              decodingMethod,
+              maxActivePaths);
+      streamObj = rcgOjb.createStream();
+    } catch (Exception e) {
+      System.err.println(e);
+      e.printStackTrace();
+    }
+  }
+
+  public void initModelWithCfg(String cfgFile) {
+    try {
+      // you should set setCfgPath() before running this
+      rcgOjb = new OnlineRecognizer(cfgFile);
+      streamObj = rcgOjb.createStream();
+    } catch (Exception e) {
+      System.err.println(e);
+      e.printStackTrace();
+    }
+  }
+
+  public void simpleExample() {
+    try {
+      float[] buffer = rcgOjb.readWavFile(wavfilename); // read data from file
+      streamObj.acceptWaveform(buffer); // feed stream with data
+      streamObj.inputFinished(); // tell engine you done with all data
+      OnlineStream ssObj[] = new OnlineStream[1];
+      while (rcgOjb.isReady(streamObj)) { // engine is ready for unprocessed data
+        ssObj[0] = streamObj;
+        rcgOjb.decodeStreams(ssObj); // decode for multiple stream
+        // rcgOjb.DecodeStream(streamObj);   // decode for single stream
+      }
+
+      String recText = "simple:" + rcgOjb.getResult(streamObj) + "\n";
+      byte[] utf8Data = recText.getBytes(StandardCharsets.UTF_8);
+      System.out.println(new String(utf8Data));
+      rcgOjb.reSet(streamObj);
+      rcgOjb.releaseStream(streamObj); // release stream
+      rcgOjb.release(); // release recognizer
+
+    } catch (Exception e) {
+      System.err.println(e);
+      e.printStackTrace();
+    }
+  }
+
+  public void streamExample() {
+    try {
+      float[] buffer = rcgOjb.readWavFile(wavfilename); // read data from file
+      float[] chunk = new float[1600]; // //each time read 1600(0.1s) data
+      int chunkIndex = 0;
+      for (int i = 0; i < buffer.length; i++) // total wav length loop
+      {
+        chunk[chunkIndex] = buffer[i];
+        chunkIndex++;
+        if (chunkIndex >= 1600 || i == (buffer.length - 1)) {
+          chunkIndex = 0;
+          streamObj.acceptWaveform(chunk); // feed chunk
+          if (rcgOjb.isReady(streamObj)) {
+            rcgOjb.decodeStream(streamObj);
+          }
+          String testDate = rcgOjb.getResult(streamObj);
+          byte[] utf8Data = testDate.getBytes(StandardCharsets.UTF_8);
+
+          if (utf8Data.length > 0) {
+            System.out.println(Float.valueOf((float) i / 16000) + ":" + new String(utf8Data));
+          }
+        }
+      }
+      streamObj.inputFinished();
+      while (rcgOjb.isReady(streamObj)) {
+        rcgOjb.decodeStream(streamObj);
+      }
+
+      String recText = "stream:" + rcgOjb.getResult(streamObj) + "\n";
+      byte[] utf8Data = recText.getBytes(StandardCharsets.UTF_8);
+      System.out.println(new String(utf8Data));
+      rcgOjb.reSet(streamObj);
+      rcgOjb.releaseStream(streamObj); // release stream
+      rcgOjb.release(); // release recognizer
+
+    } catch (Exception e) {
+      System.err.println(e);
+      e.printStackTrace();
+    }
+  }
+
+  public static void main(String[] args) {
+    try {
+      String appDir = System.getProperty("user.dir");
+      System.out.println("appdir=" + appDir);
+      String fileName = appDir + "/test.wav";
+      String cfgPath = appDir + "/modelconfig.cfg";
+      String soPath = appDir + "/../build/lib/libsherpa-onnx-jni.so";
+      OnlineRecognizer.setSoPath(soPath);
+      DecodeFile rcgDemo = new DecodeFile(fileName);
+
+      // ***************** */
+      rcgDemo.initModelWithCfg(cfgPath);
+      rcgDemo.streamExample();
+      // **************** */
+      rcgDemo.initModelWithCfg(cfgPath);
+      rcgDemo.simpleExample();
+
+    } catch (Exception e) {
+      System.err.println(e);
+      e.printStackTrace();
+    }
+  }
+}