Add Kotlin and Java API for Dolphin CTC models (#2086)

java-api-examples/NonStreamingDecodeFileDolphinCtc.java

@@ -0,0 +1,49 @@
+// Copyright 2025 Xiaomi Corporation
+
+// This file shows how to use an offline Dolphin CTC model, i.e.,
+// non-streaming Dolphin CTC model, to decode files.
+import com.k2fsa.sherpa.onnx.*;
+
+public class NonStreamingDecodeFileDolphinCtc {
+  public static void main(String[] args) {
+    // please refer to
+    // https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
+    // to download model files
+    String model = "./sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02/model.int8.onnx";
+    String tokens = "./sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02/tokens.txt";
+
+    String waveFilename =
+        "./sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02/test_wavs/0.wav";
+
+    WaveReader reader = new WaveReader(waveFilename);
+
+    OfflineDolphinModelConfig dolphin = OfflineDolphinModelConfig.builder().setModel(model).build();
+
+    OfflineModelConfig modelConfig =
+        OfflineModelConfig.builder()
+            .setDolphin(dolphin)
+            .setTokens(tokens)
+            .setNumThreads(1)
+            .setDebug(true)
+            .build();
+
+    OfflineRecognizerConfig config =
+        OfflineRecognizerConfig.builder()
+            .setOfflineModelConfig(modelConfig)
+            .setDecodingMethod("greedy_search")
+            .build();
+
+    OfflineRecognizer recognizer = new OfflineRecognizer(config);
+    OfflineStream stream = recognizer.createStream();
+    stream.acceptWaveform(reader.getSamples(), reader.getSampleRate());
+
+    recognizer.decode(stream);
+
+    String text = recognizer.getResult(stream).getText();
+
+    System.out.printf("filename:%s\nresult:%s\n", waveFilename, text);
+
+    stream.release();
+    recognizer.release();
+  }
+}

java-api-examples/README.md

@@ -23,6 +23,7 @@ This directory contains examples for the JAVA API of sherpa-onnx.
 ## Non-Streaming Speech recognition
 
 ```bash
+./run-non-streaming-decode-file-dolphin-ctc.sh
 ./run-non-streaming-decode-file-paraformer.sh
 ./run-non-streaming-decode-file-sense-voice.sh
 ./run-non-streaming-decode-file-transducer.sh
@@ -102,6 +103,12 @@ The punctuation model supports both English and Chinese.
 ./run-vad-remove-slience.sh
 ```
 
+## VAD + Non-streaming Dolphin CTC for speech recognition
+
+```bash
+./run-vad-non-streaming-dolphin-ctc.sh
+```
+
 ## VAD + Non-streaming SenseVoice for speech recognition
 
 ```bash

java-api-examples/VadNonStreamingDolphinCtc.java

@@ -0,0 +1,123 @@
+// Copyright 2025 Xiaomi Corporation
+
+// This file shows how to use a silero_vad model with a non-streaming Dolphin
+// CTC model for speech recognition.
+
+import com.k2fsa.sherpa.onnx.*;
+import java.util.Arrays;
+
+public class VadNonStreamingSenseVoice {
+  public static Vad createVad() {
+    // please download ./silero_vad.onnx from
+    // https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
+    String model = "./silero_vad.onnx";
+    SileroVadModelConfig sileroVad =
+        SileroVadModelConfig.builder()
+            .setModel(model)
+            .setThreshold(0.5f)
+            .setMinSilenceDuration(0.25f)
+            .setMinSpeechDuration(0.5f)
+            .setWindowSize(512)
+            .setMaxSpeechDuration(5.0f)
+            .build();
+
+    VadModelConfig config =
+        VadModelConfig.builder()
+            .setSileroVadModelConfig(sileroVad)
+            .setSampleRate(16000)
+            .setNumThreads(1)
+            .setDebug(true)
+            .setProvider("cpu")
+            .build();
+
+    return new Vad(config);
+  }
+
+  public static OfflineRecognizer createOfflineRecognizer() {
+    // please refer to
+    // https://k2-fsa.github.io/sherpa/onnx/dolphin/index.html
+    // to download model files
+    String model = "./sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02/model.int8.onnx";
+    String tokens = "./sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02/tokens.txt";
+
+    OfflineDolphinModelConfig dolphin = OfflineDolphinModelConfig.builder().setModel(model).build();
+
+    OfflineModelConfig modelConfig =
+        OfflineModelConfig.builder()
+            .setDolphin(dolphin)
+            .setTokens(tokens)
+            .setNumThreads(1)
+            .setDebug(true)
+            .build();
+
+    OfflineRecognizerConfig config =
+        OfflineRecognizerConfig.builder()
+            .setOfflineModelConfig(modelConfig)
+            .setDecodingMethod("greedy_search")
+            .build();
+
+    return new OfflineRecognizer(config);
+  }
+
+  public static void main(String[] args) {
+
+    Vad vad = createVad();
+    OfflineRecognizer recognizer = createOfflineRecognizer();
+
+    // You can download the test file from
+    // https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
+    String testWaveFilename = "./lei-jun-test.wav";
+    WaveReader reader = new WaveReader(testWaveFilename);
+
+    int numSamples = reader.getSamples().length;
+    int numIter = numSamples / 512;
+
+    for (int i = 0; i != numIter; ++i) {
+      int start = i * 512;
+      int end = start + 512;
+      float[] samples = Arrays.copyOfRange(reader.getSamples(), start, end);
+      vad.acceptWaveform(samples);
+      if (vad.isSpeechDetected()) {
+        while (!vad.empty()) {
+          SpeechSegment segment = vad.front();
+          float startTime = segment.getStart() / 16000.0f;
+          float duration = segment.getSamples().length / 16000.0f;
+
+          OfflineStream stream = recognizer.createStream();
+          stream.acceptWaveform(segment.getSamples(), 16000);
+          recognizer.decode(stream);
+          String text = recognizer.getResult(stream).getText();
+          stream.release();
+
+          if (!text.isEmpty()) {
+            System.out.printf("%.3f--%.3f: %s\n", startTime, startTime + duration, text);
+          }
+
+          vad.pop();
+        }
+      }
+    }
+
+    vad.flush();
+    while (!vad.empty()) {
+      SpeechSegment segment = vad.front();
+      float startTime = segment.getStart() / 16000.0f;
+      float duration = segment.getSamples().length / 16000.0f;
+
+      OfflineStream stream = recognizer.createStream();
+      stream.acceptWaveform(segment.getSamples(), 16000);
+      recognizer.decode(stream);
+      String text = recognizer.getResult(stream).getText();
+      stream.release();
+
+      if (!text.isEmpty()) {
+        System.out.printf("%.3f--%.3f: %s\n", startTime, startTime + duration, text);
+      }
+
+      vad.pop();
+    }
+
+    vad.release();
+    recognizer.release();
+  }
+}

java-api-examples/run-non-streaming-decode-file-dolphin-ctc.sh

@@ -0,0 +1,38 @@
+#!/usr/bin/env bash
+
+set -ex
+
+if [[ ! -f ../build/lib/libsherpa-onnx-jni.dylib  && ! -f ../build/lib/libsherpa-onnx-jni.so ]]; then
+  mkdir -p ../build
+  pushd ../build
+  cmake \
+    -DSHERPA_ONNX_ENABLE_PYTHON=OFF \
+    -DSHERPA_ONNX_ENABLE_TESTS=OFF \
+    -DSHERPA_ONNX_ENABLE_CHECK=OFF \
+    -DBUILD_SHARED_LIBS=ON \
+    -DSHERPA_ONNX_ENABLE_PORTAUDIO=OFF \
+    -DSHERPA_ONNX_ENABLE_JNI=ON \
+    ..
+
+  make -j4
+  ls -lh lib
+  popd
+fi
+
+if [ ! -f ../sherpa-onnx/java-api/build/sherpa-onnx.jar ]; then
+  pushd ../sherpa-onnx/java-api
+  make
+  popd
+fi
+
+if [ ! -f ./sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02/model.int8.onnx ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02.tar.bz2
+  tar xvf sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02.tar.bz2
+  rm sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02.tar.bz2
+  ls -lh sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02
+fi
+
+java \
+  -Djava.library.path=$PWD/../build/lib \
+  -cp ../sherpa-onnx/java-api/build/sherpa-onnx.jar \
+  NonStreamingDecodeFileDolphinCtc.java

java-api-examples/run-vad-non-streaming-dolphin-ctc.sh

@@ -0,0 +1,46 @@
+#!/usr/bin/env bash
+
+set -ex
+
+if [[ ! -f ../build/lib/libsherpa-onnx-jni.dylib  && ! -f ../build/lib/libsherpa-onnx-jni.so ]]; then
+  mkdir -p ../build
+  pushd ../build
+  cmake \
+    -DSHERPA_ONNX_ENABLE_PYTHON=OFF \
+    -DSHERPA_ONNX_ENABLE_TESTS=OFF \
+    -DSHERPA_ONNX_ENABLE_CHECK=OFF \
+    -DBUILD_SHARED_LIBS=ON \
+    -DSHERPA_ONNX_ENABLE_PORTAUDIO=OFF \
+    -DSHERPA_ONNX_ENABLE_JNI=ON \
+    ..
+
+  make -j4
+  ls -lh lib
+  popd
+fi
+
+if [ ! -f ../sherpa-onnx/java-api/build/sherpa-onnx.jar ]; then
+  pushd ../sherpa-onnx/java-api
+  make
+  popd
+fi
+
+if [ ! -f ./silero_vad.onnx ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx
+fi
+
+if [ ! -f ./lei-jun-test.wav ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/lei-jun-test.wav
+fi
+
+if [ ! -f ./sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02/model.int8.onnx ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02.tar.bz2
+  tar xvf sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02.tar.bz2
+  rm sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02.tar.bz2
+  ls -lh sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02
+fi
+
+java \
+  -Djava.library.path=$PWD/../build/lib \
+  -cp ../sherpa-onnx/java-api/build/sherpa-onnx.jar \
+  ./VadNonStreamingDolphinCtc.java