Add Java API for audio tagging (#820)

@@ -106,6 +106,17 @@ jobs:
             make -j4
             ls -lh lib
 
+      - name:  Run java test (audio tagging)
+        shell: bash
+        run: |
+          cd ./java-api-examples
+          ./run-audio-tagging-zipformer-from-file.sh
+          # Delete model files to save space
+          rm -rf sherpa-onnx-zipformer-*
+
+          ./run-audio-tagging-ced-from-file.sh
+          rm -rf sherpa-onnx-ced-*
+
       - name:  Run java test (add punctuations)
         shell: bash
         run: |

@@ -101,3 +101,4 @@ sherpa-onnx-kws-zipformer-wenetspeech-3.3M-2024-01-01
 *.tar.gz
 *.tar.bz2
 *.zip
+sherpa-onnx-ced-*

+// Copyright 2024 Xiaomi Corporation
+
+// This file shows how to use a CED audio tagging model to tag
+// input audio files.
+import com.k2fsa.sherpa.onnx.*;
+
+public class AudioTaggingCEDFromFile {
+  public static void main(String[] args) {
+    // please download the model from
+    // https://github.com/k2-fsa/sherpa-onnx/releases/tag/audio-tagging-models
+    String model = "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/model.int8.onnx";
+    String labels = "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/class_labels_indices.csv";
+    int topK = 5;
+
+    String[] testWaves =
+        new String[] {
+          "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/test_wavs/1.wav",
+          "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/test_wavs/2.wav",
+          "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/test_wavs/3.wav",
+          "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/test_wavs/4.wav",
+          "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/test_wavs/5.wav",
+          "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/test_wavs/6.wav",
+          "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/test_wavs/7.wav",
+          "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/test_wavs/8.wav",
+          "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/test_wavs/9.wav",
+          "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/test_wavs/10.wav",
+          "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/test_wavs/11.wav",
+          "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/test_wavs/12.wav",
+          "./sherpa-onnx-ced-mini-audio-tagging-2024-04-19/test_wavs/13.wav",
+        };
+
+    AudioTaggingModelConfig modelConfig =
+        AudioTaggingModelConfig.builder().setCED(model).setNumThreads(1).setDebug(true).build();
+
+    AudioTaggingConfig config =
+        AudioTaggingConfig.builder().setModel(modelConfig).setLabels(labels).setTopK(topK).build();
+
+    AudioTagging tagger = new AudioTagging(config);
+    System.out.println("------");
+    for (String filename : testWaves) {
+      WaveReader reader = new WaveReader(filename);
+
+      OfflineStream stream = tagger.createStream();
+      stream.acceptWaveform(reader.getSamples(), reader.getSampleRate());
+
+      AudioEvent[] events = tagger.compute(stream);
+
+      stream.release();
+
+      System.out.printf("input file: %s\n", filename);
+      System.out.printf("Probability\t\tName\n");
+      for (AudioEvent e : events) {
+        System.out.printf("%.3f\t\t\t%s\n", e.getProb(), e.getName());
+      }
+      System.out.println("------");
+    }
+
+    tagger.release();
+  }
+}

+// Copyright 2024 Xiaomi Corporation
+
+// This file shows how to use a zipformer audio tagging model to tag
+// input audio files.
+import com.k2fsa.sherpa.onnx.*;
+
+public class AudioTaggingZipformerFromFile {
+  public static void main(String[] args) {
+    // please download the model from
+    // https://github.com/k2-fsa/sherpa-onnx/releases/tag/audio-tagging-models
+    String model = "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/model.int8.onnx";
+    String labels =
+        "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/class_labels_indices.csv";
+    int topK = 5;
+
+    String[] testWaves =
+        new String[] {
+          "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/test_wavs/1.wav",
+          "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/test_wavs/2.wav",
+          "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/test_wavs/3.wav",
+          "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/test_wavs/4.wav",
+          "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/test_wavs/5.wav",
+          "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/test_wavs/6.wav",
+          "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/test_wavs/7.wav",
+          "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/test_wavs/8.wav",
+          "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/test_wavs/9.wav",
+          "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/test_wavs/10.wav",
+          "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/test_wavs/11.wav",
+          "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/test_wavs/12.wav",
+          "./sherpa-onnx-zipformer-small-audio-tagging-2024-04-15/test_wavs/13.wav",
+        };
+
+    OfflineZipformerAudioTaggingModelConfig zipformer =
+        OfflineZipformerAudioTaggingModelConfig.builder().setModel(model).build();
+
+    AudioTaggingModelConfig modelConfig =
+        AudioTaggingModelConfig.builder()
+            .setZipformer(zipformer)
+            .setNumThreads(1)
+            .setDebug(true)
+            .build();
+
+    AudioTaggingConfig config =
+        AudioTaggingConfig.builder().setModel(modelConfig).setLabels(labels).setTopK(topK).build();
+
+    AudioTagging tagger = new AudioTagging(config);
+    System.out.println("------");
+    for (String filename : testWaves) {
+      WaveReader reader = new WaveReader(filename);
+
+      OfflineStream stream = tagger.createStream();
+      stream.acceptWaveform(reader.getSamples(), reader.getSampleRate());
+
+      AudioEvent[] events = tagger.compute(stream);
+
+      stream.release();
+
+      System.out.printf("input file: %s\n", filename);
+      System.out.printf("Probability\t\tName\n");
+      for (AudioEvent e : events) {
+        System.out.printf("%.3f\t\t\t%s\n", e.getProb(), e.getName());
+      }
+      System.out.println("------");
+    }
+
+    tagger.release();
+  }
+}

@@ -36,10 +36,17 @@ This directory contains examples for the JAVA API of sherpa-onnx.
 ./run-spoken-language-identification-whisper.sh
 ```
 
-## Add puncutations to text
+## Add punctuations to text
 
 The punctuation model supports both English and Chinese.
 
 ```bash
 ./run-add-punctuation-zh-en.sh
 ```
+
+## Audio tagging
+
+```bash
+./run-audio-tagging-zipformer-from-file.sh
+./run-audio-tagging-ced-from-file.sh
+```

@@ -25,20 +25,6 @@ if [ ! -f ../sherpa-onnx/java-api/build/sherpa-onnx.jar ]; then
   popd
 fi
 
-if [[ ! -f ../build/lib/libsherpa-onnx-jni.dylib  && ! -f ../build/lib/libsherpa-onnx-jni.so ]]; then
-  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
-fi
-
 if [ ! -f ./sherpa-onnx-punct-ct-transformer-zh-en-vocab272727-2024-04-12/model.onnx ]; then
   curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/punctuation-models/sherpa-onnx-punct-ct-transformer-zh-en-vocab272727-2024-04-12.tar.bz2
   tar xvf sherpa-onnx-punct-ct-transformer-zh-en-vocab272727-2024-04-12.tar.bz2

+#!/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-ced-mini-audio-tagging-2024-04-19/model.int8.onnx ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/audio-tagging-models/sherpa-onnx-ced-mini-audio-tagging-2024-04-19.tar.bz2
+  tar xvf sherpa-onnx-ced-mini-audio-tagging-2024-04-19.tar.bz2
+  rm sherpa-onnx-ced-mini-audio-tagging-2024-04-19.tar.bz2
+fi
+
+java \
+  -Djava.library.path=$PWD/../build/lib \
+  -cp ../sherpa-onnx/java-api/build/sherpa-onnx.jar \
+  ./AudioTaggingCEDFromFile.java

+#!/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-zipformer-small-audio-tagging-2024-04-15/model.int8.onnx ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/audio-tagging-models/sherpa-onnx-zipformer-small-audio-tagging-2024-04-15.tar.bz2
+  tar xvf sherpa-onnx-zipformer-small-audio-tagging-2024-04-15.tar.bz2
+  rm sherpa-onnx-zipformer-small-audio-tagging-2024-04-15.tar.bz2
+fi
+
+java \
+  -Djava.library.path=$PWD/../build/lib \
+  -cp ../sherpa-onnx/java-api/build/sherpa-onnx.jar \
+  ./AudioTaggingZipformerFromFile.java

@@ -25,20 +25,6 @@ if [ ! -f ../sherpa-onnx/java-api/build/sherpa-onnx.jar ]; then
   popd
 fi
 
-if [[ ! -f ../build/lib/libsherpa-onnx-jni.dylib  && ! -f ../build/lib/libsherpa-onnx-jni.so ]]; then
-  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
-fi
-
 if [ ! -f ./sherpa-onnx-nemo-ctc-en-citrinet-512/tokens.txt ]; then
   curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-nemo-ctc-en-citrinet-512.tar.bz2
   tar xvf sherpa-onnx-nemo-ctc-en-citrinet-512.tar.bz2

@@ -25,20 +25,6 @@ if [ ! -f ../sherpa-onnx/java-api/build/sherpa-onnx.jar ]; then
   popd
 fi
 
-if [[ ! -f ../build/lib/libsherpa-onnx-jni.dylib  && ! -f ../build/lib/libsherpa-onnx-jni.so ]]; then
-  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
-fi
-
 if [ ! -f ./sherpa-onnx-paraformer-zh-2023-03-28/tokens.txt ]; then
   curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
 

@@ -25,20 +25,6 @@ if [ ! -f ../sherpa-onnx/java-api/build/sherpa-onnx.jar ]; then
   popd
 fi
 
-if [[ ! -f ../build/lib/libsherpa-onnx-jni.dylib  && ! -f ../build/lib/libsherpa-onnx-jni.so ]]; then
-  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
-fi
-
 if [ ! -f ./sherpa-onnx-zipformer-gigaspeech-2023-12-12/tokens.txt ]; then
   curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-zipformer-gigaspeech-2023-12-12.tar.bz2
 

@@ -25,20 +25,6 @@ if [ ! -f ../sherpa-onnx/java-api/build/sherpa-onnx.jar ]; then
   popd
 fi
 
-if [[ ! -f ../build/lib/libsherpa-onnx-jni.dylib  && ! -f ../build/lib/libsherpa-onnx-jni.so ]]; then
-  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
-fi
-
 if [ ! -f ./sherpa-onnx-whisper-tiny.en/tiny.en-tokens.txt ]; then
   curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-whisper-tiny.en.tar.bz2
 

@@ -25,20 +25,6 @@ if [ ! -f ../sherpa-onnx/java-api/build/sherpa-onnx.jar ]; then
   popd
 fi
 
-if [[ ! -f ../build/lib/libsherpa-onnx-jni.dylib  && ! -f ../build/lib/libsherpa-onnx-jni.so ]]; then
-  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
-fi
-
 # please visit
 # https://github.com/k2-fsa/sherpa-onnx/releases/tag/tts-models
 # to download more models

@@ -25,20 +25,6 @@ if [ ! -f ../sherpa-onnx/java-api/build/sherpa-onnx.jar ]; then
   popd
 fi
 
-if [[ ! -f ../build/lib/libsherpa-onnx-jni.dylib  && ! -f ../build/lib/libsherpa-onnx-jni.so ]]; then
-  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
-fi
-
 # please visit
 # https://github.com/k2-fsa/sherpa-onnx/releases/tag/tts-models
 # to download more models

@@ -25,20 +25,6 @@ if [ ! -f ../sherpa-onnx/java-api/build/sherpa-onnx.jar ]; then
   popd
 fi
 
-if [[ ! -f ../build/lib/libsherpa-onnx-jni.dylib  && ! -f ../build/lib/libsherpa-onnx-jni.so ]]; then
-  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
-fi
-
 # please visit
 # https://github.com/k2-fsa/sherpa-onnx/releases/tag/tts-models
 # to download more models

@@ -25,20 +25,6 @@ if [ ! -f ../sherpa-onnx/java-api/build/sherpa-onnx.jar ]; then
   popd
 fi
 
-if [[ ! -f ../build/lib/libsherpa-onnx-jni.dylib  && ! -f ../build/lib/libsherpa-onnx-jni.so ]]; then
-  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
-fi
-
 # Note that it needs a multilingual whisper model. so, for example, tiny works while tiny.en does not work
 # https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-whisper-tiny.tar.bz2
 if [ ! -f ./sherpa-onnx-whisper-tiny/tiny-encoder.int8.onnx ]; then

+# Copyright 2024 Xiaomi Corporation
 
 # all .class and .jar files are put inside out_dir
 out_dir := build
@@ -44,6 +45,12 @@ java_files += OfflinePunctuationModelConfig.java
 java_files += OfflinePunctuationConfig.java
 java_files += OfflinePunctuation.java
 
+java_files += OfflineZipformerAudioTaggingModelConfig.java
+java_files += AudioTaggingModelConfig.java
+java_files += AudioTaggingConfig.java
+java_files += AudioEvent.java
+java_files += AudioTagging.java
+
 class_files := $(java_files:%.java=%.class)
 
 java_files := $(addprefix src/$(package_dir)/,$(java_files))

+// Copyright 2024 Xiaomi Corporation
+
+package com.k2fsa.sherpa.onnx;
+
+public class AudioEvent {
+    private String name = "";
+    private int index = 0;
+    private float prob = 0;
+
+    public AudioEvent(String name, int index, float prob) {
+        this.name = name;
+        this.index = index;
+        this.prob = prob;
+    }
+
+    public String getName() {
+        return name;
+    }
+
+    public int getIndex() {
+        return index;
+    }
+
+    public float getProb() {
+        return prob;
+    }
+
+    @Override
+    public String toString() {
+        return String.format("AudioEven(name=%s, index=%d, prob=%.3f)\n", name, index, prob);
+    }
+}

+// Copyright 2024 Xiaomi Corporation
+
+package com.k2fsa.sherpa.onnx;
+
+public class AudioTagging {
+    static {
+        System.loadLibrary("sherpa-onnx-jni");
+    }
+
+    private long ptr = 0;
+
+    public AudioTagging(AudioTaggingConfig config) {
+        ptr = newFromFile(config);
+    }
+
+    public OfflineStream createStream() {
+        long p = createStream(ptr);
+        return new OfflineStream(p);
+    }
+
+    public AudioEvent[] compute(OfflineStream stream) {
+        return compute(stream, -1);
+
+    }
+
+    public AudioEvent[] compute(OfflineStream stream, int topK) {
+        Object[] arr = compute(ptr, stream.getPtr(), topK);
+
+        AudioEvent[] events = new AudioEvent[arr.length];
+        for (int i = 0; i < arr.length; ++i) {
+            Object[] obj = (Object[]) arr[i];
+            String name = (String) obj[0];
+            int index = (int) obj[1];
+            float prob = (float) obj[2];
+            events[i] = new AudioEvent(name, index, prob);
+        }
+        return events;
+    }
+
+
+    @Override
+    protected void finalize() throws Throwable {
+        release();
+    }
+
+    // You'd better call it manually if it is not used anymore
+    public void release() {
+        if (this.ptr == 0) {
+            return;
+        }
+        delete(this.ptr);
+        this.ptr = 0;
+    }
+
+    private native void delete(long ptr);
+
+    private native long newFromFile(AudioTaggingConfig config);
+
+    private native long createStream(long ptr);
+
+    private native Object[] compute(long ptr, long streamPtr, int topK);
+}

+// Copyright 2024 Xiaomi Corporation
+
+package com.k2fsa.sherpa.onnx;
+
+public class AudioTaggingConfig {
+    private final AudioTaggingModelConfig model;
+    private final String labels;
+    private final int topK;
+
+    private AudioTaggingConfig(Builder builder) {
+        this.model = builder.model;
+        this.labels = builder.labels;
+        this.topK = builder.topK;
+    }
+
+    public static Builder builder() {
+        return new AudioTaggingConfig.Builder();
+    }
+
+    public static class Builder {
+        private AudioTaggingModelConfig model = AudioTaggingModelConfig.builder().build();
+        private String labels = "";
+        private int topK = 5;
+
+        public AudioTaggingConfig build() {
+            return new AudioTaggingConfig(this);
+        }
+
+        public Builder setModel(AudioTaggingModelConfig model) {
+            this.model = model;
+            return this;
+        }
+
+        public Builder setLabels(String labels) {
+            this.labels = labels;
+            return this;
+        }
+
+        public Builder setTopK(int topK) {
+            this.topK = topK;
+            return this;
+        }
+    }
+}

+// Copyright 2024 Xiaomi Corporation
+
+package com.k2fsa.sherpa.onnx;
+
+public class AudioTaggingModelConfig {
+    private final OfflineZipformerAudioTaggingModelConfig zipformer;
+    private final String ced;
+    private final int numThreads;
+    private final boolean debug;
+    private final String provider;
+
+    private AudioTaggingModelConfig(Builder builder) {
+        this.zipformer = builder.zipformer;
+        this.ced = builder.ced;
+        this.numThreads = builder.numThreads;
+        this.debug = builder.debug;
+        this.provider = builder.provider;
+    }
+
+    public static Builder builder() {
+        return new Builder();
+    }
+
+    public static class Builder {
+        private OfflineZipformerAudioTaggingModelConfig zipformer = OfflineZipformerAudioTaggingModelConfig.builder().build();
+        private String ced = "";
+        private int numThreads = 1;
+        private boolean debug = true;
+        private String provider = "cpu";
+
+        public AudioTaggingModelConfig build() {
+            return new AudioTaggingModelConfig(this);
+        }
+
+        public Builder setZipformer(OfflineZipformerAudioTaggingModelConfig zipformer) {
+            this.zipformer = zipformer;
+            return this;
+        }
+
+        public Builder setCED(String ced) {
+            this.ced = ced;
+            return this;
+        }
+
+        public Builder setNumThreads(int numThreads) {
+            this.numThreads = numThreads;
+            return this;
+        }
+
+        public Builder setDebug(boolean debug) {
+            this.debug = debug;
+            return this;
+        }
+
+        public Builder setProvider(String provider) {
+            this.provider = provider;
+            return this;
+        }
+    }
+}

 // Copyright 2022-2023 by zhaoming
 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class EndpointRule {

 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OfflineModelConfig {

 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OfflineNemoEncDecCtcModelConfig {

 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OfflineParaformerModelConfig {

@@ -7,7 +7,7 @@ public class OfflinePunctuation {
         System.loadLibrary("sherpa-onnx-jni");
     }
 
-    private long ptr = 0; // this is the asr engine ptrss
+    private long ptr = 0;
 
     public OfflinePunctuation(OfflinePunctuationConfig config) {
         ptr = newFromFile(config);

 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OfflineRecognizer {

 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OfflineRecognizerConfig {

 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OfflineRecognizerResult {

 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OfflineStream {

 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OfflineTransducerModelConfig {

 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OfflineWhisperModelConfig {

+// Copyright 2024 Xiaomi Corporation
+
+package com.k2fsa.sherpa.onnx;
+
+public class OfflineZipformerAudioTaggingModelConfig {
+    private final String model;
+
+    private OfflineZipformerAudioTaggingModelConfig(Builder builder) {
+        this.model = builder.model;
+    }
+
+    public static Builder builder() {
+        return new Builder();
+    }
+
+    public String getModel() {
+        return model;
+    }
+
+    public static class Builder {
+        private String model = "";
+
+        public OfflineZipformerAudioTaggingModelConfig build() {
+            return new OfflineZipformerAudioTaggingModelConfig(this);
+        }
+
+        public Builder setModel(String model) {
+            this.model = model;
+            return this;
+        }
+    }
+}

 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OnlineCtcFstDecoderConfig {

 // Copyright 2022-2023 by zhaoming
 // Copyright 2024 Xiaomi Corporation
-package com.k2fsa.sherpa.onnx;
 
+package com.k2fsa.sherpa.onnx;
 
 public class OnlineRecognizer {
     static {

 // Copyright 2022-2023 by zhaoming
 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OnlineRecognizerConfig {

 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OnlineRecognizerResult {

 // Copyright 2022-2023 by zhaoming
 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OnlineStream {

 // Copyright 2024 Xiaomi Corporation
+
 package com.k2fsa.sherpa.onnx;
 
 public class OnlineZipformer2CtcModelConfig {