Add Kotlin and Java API for Google MedAsr model (#2956)

.github/workflows/run-java-test.yaml

@@ -108,6 +108,14 @@ jobs:
           cd ./java-api-examples
           ./run-version-test.sh
 
+      - name:  Run java test (MedASR CTC)
+        shell: bash
+        run: |
+          cd ./java-api-examples
+          ./run-non-streaming-decode-file-medasr-ctc.sh
+          rm -rf sherpa-onnx-medasr-*
+
+
       - name:  Run java test (Omnilingual ASR CTC)
         shell: bash
         run: |

java-api-examples/NonStreamingDecodeFileMedAsrCtc.java

@@ -0,0 +1,51 @@
+// Copyright 2025 Xiaomi Corporation
+
+// This file shows how to use an offline Google MedASR CTC model,
+// i.e., non-streaming MedASR CTC model,
+// to decode files.
+import com.k2fsa.sherpa.onnx.*;
+
+public class NonStreamingDecodeFileMedAsrCtc {
+  public static void main(String[] args) {
+    // please refer to
+    // https://k2-fsa.github.io/sherpa/onnx/medasr/index.html
+    // to download model files
+    String model = "./sherpa-onnx-medasr-ctc-en-int8-2025-12-25/model.int8.onnx";
+
+    String tokens = "./sherpa-onnx-medasr-ctc-en-int8-2025-12-25/tokens.txt";
+
+    String waveFilename = "./sherpa-onnx-medasr-ctc-en-int8-2025-12-25/test_wavs/0.wav";
+
+    WaveReader reader = new WaveReader(waveFilename);
+
+    OfflineMedAsrCtcModelConfig medasr =
+        OfflineMedAsrCtcModelConfig.builder().setModel(model).build();
+
+    OfflineModelConfig modelConfig =
+        OfflineModelConfig.builder()
+            .setMedAsr(medasr)
+            .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/NonStreamingDecodeFileOmnilingualAsrCtc.java

@@ -8,7 +8,7 @@ import com.k2fsa.sherpa.onnx.*;
 public class NonStreamingDecodeFileOmnilingualAsrCtc {
   public static void main(String[] args) {
     // please refer to
-    // https://k2-fsa.github.io/sherpa/onnx/sense-voice/index.html
+    // https://k2-fsa.github.io/sherpa/onnx/omnilingual-asr/index.html
     // to download model files
     String model =
         "sherpa-onnx-omnilingual-asr-1600-languages-300M-ctc-int8-2025-11-12/model.int8.onnx";

java-api-examples/run-non-streaming-decode-file-medasr-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-medasr-ctc-en-int8-2025-12-25/tokens.txt ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-medasr-ctc-en-int8-2025-12-25.tar.bz2
+  tar xvf sherpa-onnx-medasr-ctc-en-int8-2025-12-25.tar.bz2
+  rm sherpa-onnx-medasr-ctc-en-int8-2025-12-25.tar.bz2
+fi
+
+
+java \
+  -Djava.library.path=$PWD/../build/lib \
+  -cp ../sherpa-onnx/java-api/build/sherpa-onnx.jar \
+  NonStreamingDecodeFileMedAsrCtc.java

kotlin-api-examples/run.sh

@@ -234,10 +234,10 @@ function testOfflineAsr() {
     rm sherpa-onnx-moonshine-tiny-en-int8.tar.bz2
   fi
 
-  if [ ! -f ./sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17/tokens.txt ]; then
-    curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17.tar.bz2
-    tar xvf sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17.tar.bz2
-    rm sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17.tar.bz2
+  if [ ! -f ./sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17/tokens.txt ]; then
+    curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17.tar.bz2
+    tar xvf sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17.tar.bz2
+    rm sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17.tar.bz2
   fi
 
   if [ ! -f ./sherpa-onnx-whisper-tiny.en/tiny.en-encoder.int8.onnx ]; then
@@ -439,10 +439,10 @@ function testOfflineSpeechDenoiser() {
 }
 
 function testOfflineSenseVoiceWithHr() {
-  if [ ! -f ./sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17/tokens.txt ]; then
-    curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17.tar.bz2
-    tar xvf sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17.tar.bz2
-    rm sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17.tar.bz2
+  if [ ! -f ./sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17/tokens.txt ]; then
+    curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17.tar.bz2
+    tar xvf sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17.tar.bz2
+    rm sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17.tar.bz2
   fi
 
   if [ ! -d dict ]; then
@@ -515,6 +515,28 @@ function testOfflineOmnilingualAsrCtc() {
   java -Djava.library.path=../build/lib -jar $out_filename
 }
 
+function testOfflineMedAsrCtc() {
+  if [ ! -f ./sherpa-onnx-medasr-ctc-en-int8-2025-12-25/tokens.txt ]; then
+    curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-medasr-ctc-en-int8-2025-12-25.tar.bz2
+    tar xvf sherpa-onnx-medasr-ctc-en-int8-2025-12-25.tar.bz2
+    rm sherpa-onnx-medasr-ctc-en-int8-2025-12-25.tar.bz2
+  fi
+
+  out_filename=test_offline_medasr_ctc.jar
+  kotlinc-jvm -include-runtime -d $out_filename \
+    test_offline_medasr_ctc.kt \
+    FeatureConfig.kt \
+    QnnConfig.kt \
+    HomophoneReplacerConfig.kt \
+    OfflineRecognizer.kt \
+    OfflineStream.kt \
+    WaveReader.kt \
+    faked-asset-manager.kt
+
+  ls -lh $out_filename
+  java -Djava.library.path=../build/lib -jar $out_filename
+}
+
 function testOfflineWenetCtc() {
   if [ ! -f sherpa-onnx-wenetspeech-yue-u2pp-conformer-ctc-zh-en-cantonese-int8-2025-09-10/model.int8.onnx ]; then
     curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-wenetspeech-yue-u2pp-conformer-ctc-zh-en-cantonese-int8-2025-09-10.tar.bz2
@@ -539,6 +561,7 @@ function testOfflineWenetCtc() {
 
 testVersion
 
+testOfflineMedAsrCtc
 testOfflineOmnilingualAsrCtc
 testOfflineWenetCtc
 testOfflineNeMoCanary

kotlin-api-examples/test_offline_asr.kt

@@ -15,7 +15,7 @@ fun test(type: Int) {
     2 -> "./sherpa-onnx-whisper-tiny.en/test_wavs/0.wav"
     5 -> "./sherpa-onnx-zipformer-multi-zh-hans-2023-9-2/test_wavs/1.wav"
     6 -> "./sherpa-onnx-nemo-ctc-en-citrinet-512/test_wavs/8k.wav"
-    15 -> "./sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17/test_wavs/zh.wav"
+    15 -> "./sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17/test_wavs/zh.wav"
     21 -> "./sherpa-onnx-moonshine-tiny-en-int8/test_wavs/0.wav"
     24 -> "./sherpa-onnx-fire-red-asr-large-zh_en-2025-02-16/test_wavs/0.wav"
     25 -> "./sherpa-onnx-dolphin-base-ctc-multi-lang-int8-2025-04-02/test_wavs/0.wav"

kotlin-api-examples/test_offline_medasr_ctc.kt

@@ -0,0 +1,31 @@
+package com.k2fsa.sherpa.onnx
+
+fun main() {
+  val recognizer = createOfflineRecognizer()
+  val waveFilename = "./sherpa-onnx-medasr-ctc-en-int8-2025-12-25/test_wavs/0.wav"
+
+  val objArray = WaveReader.readWaveFromFile(
+      filename = waveFilename,
+  )
+  val samples: FloatArray = objArray[0] as FloatArray
+  val sampleRate: Int = objArray[1] as Int
+
+  var stream = recognizer.createStream()
+  stream.acceptWaveform(samples, sampleRate=sampleRate)
+  recognizer.decode(stream)
+
+  var result = recognizer.getResult(stream)
+  println(result)
+
+  stream.release()
+  recognizer.release()
+}
+
+
+fun createOfflineRecognizer(): OfflineRecognizer {
+  val config = OfflineRecognizerConfig(
+      modelConfig = getOfflineModelConfig(type = 45)!!,
+  )
+
+  return OfflineRecognizer(config = config)
+}

nodejs-examples/README.md

@@ -287,9 +287,9 @@ how to decode a file with a non-streaming SenseVoice model with homophone replac
 You can use the following command to run it:
 
 ```bash
-curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17.tar.bz2
-tar xvf sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17.tar.bz2
-rm sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17.tar.bz2
+curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17.tar.bz2
+tar xvf sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17.tar.bz2
+rm sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17.tar.bz2
 
 curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/hr-files/dict.tar.bz2
 tar xf dict.tar.bz2
@@ -309,9 +309,9 @@ how to decode a file with a non-streaming SenseVoice model.
 You can use the following command to run it:
 
 ```bash
-wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17.tar.bz2
-tar xvf sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17.tar.bz2
-rm sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17.tar.bz2
+wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17.tar.bz2
+tar xvf sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17.tar.bz2
+rm sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17.tar.bz2
 
 node ./test-offline-sense-voice.js
 ```

nodejs-examples/test-offline-sense-voice-with-hr.js

@@ -6,11 +6,12 @@ function createOfflineRecognizer() {
   let modelConfig = {
     senseVoice: {
       model:
-          './sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17/model.int8.onnx',
+          './sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17/model.int8.onnx',
       language: '',
       useInverseTextNormalization: 1,
     },
-    tokens: './sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17/tokens.txt',
+    tokens:
+        './sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17/tokens.txt',
   };
 
   let config = {

nodejs-examples/test-offline-sense-voice.js

@@ -6,11 +6,12 @@ function createOfflineRecognizer() {
   let modelConfig = {
     senseVoice: {
       model:
-          './sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17/model.int8.onnx',
+          './sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17/model.int8.onnx',
       language: '',
       useInverseTextNormalization: 1,
     },
-    tokens: './sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17/tokens.txt',
+    tokens:
+        './sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17/tokens.txt',
   };
 
   let config = {
@@ -24,7 +25,7 @@ const recognizer = createOfflineRecognizer();
 const stream = recognizer.createStream();
 
 const waveFilename =
-    './sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17/test_wavs/zh.wav';
+    './sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17/test_wavs/zh.wav';
 const wave = sherpa_onnx.readWave(waveFilename);
 stream.acceptWaveform(wave.sampleRate, wave.samples);
 

scripts/apk/generate-vad-asr-apk-script.py

@@ -758,6 +758,22 @@ def get_models():
 
             ls -lh
 
+            popd
+            """,
+        ),
+        Model(
+            model_name="sherpa-onnx-medasr-ctc-en-int8-2025-12-25",
+            idx=45,
+            lang="en",
+            lang2="英语",
+            short_name="google_medasr_ctc_int8",
+            cmd="""
+            pushd $model_name
+
+            rm -rfv test_wavs
+
+            ls -lh
+
             popd
             """,
         ),

sherpa-onnx/java-api/Makefile

@@ -39,6 +39,7 @@ java_files += OfflineNemoEncDecCtcModelConfig.java
 java_files += OfflineZipformerCtcModelConfig.java
 java_files += OfflineWenetCtcModelConfig.java
 java_files += OfflineOmnilingualAsrCtcModelConfig.java
+java_files += OfflineMedAsrCtcModelConfig.java
 java_files += OfflineCanaryModelConfig.java
 java_files += OfflineSenseVoiceModelConfig.java
 java_files += OfflineDolphinModelConfig.java

sherpa-onnx/java-api/src/main/java/com/k2fsa/sherpa/onnx/OfflineMedAsrCtcModelConfig.java

@@ -0,0 +1,30 @@
+package com.k2fsa.sherpa.onnx;
+
+public class OfflineMedAsrCtcModelConfig {
+    private final String model;
+
+    private OfflineMedAsrCtcModelConfig(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 OfflineMedAsrCtcModelConfig build() {
+            return new OfflineMedAsrCtcModelConfig(this);
+        }
+
+        public Builder setModel(String model) {
+            this.model = model;
+            return this;
+        }
+    }
+}

sherpa-onnx/java-api/src/main/java/com/k2fsa/sherpa/onnx/OfflineModelConfig.java

@@ -14,6 +14,7 @@ public class OfflineModelConfig {
     private final OfflineZipformerCtcModelConfig zipformerCtc;
     private final OfflineWenetCtcModelConfig wenetCtc;
     private final OfflineOmnilingualAsrCtcModelConfig omnilingual;
+    private final OfflineMedAsrCtcModelConfig medasr;
     private final OfflineCanaryModelConfig canary;
     private final String teleSpeech;
     private final String tokens;
@@ -36,6 +37,7 @@ public class OfflineModelConfig {
         this.canary = builder.canary;
         this.wenetCtc = builder.wenetCtc;
         this.omnilingual = builder.omnilingual;
+        this.medasr = builder.medasr;
         this.senseVoice = builder.senseVoice;
         this.dolphin = builder.dolphin;
         this.teleSpeech = builder.teleSpeech;
@@ -92,6 +94,10 @@ public class OfflineModelConfig {
         return omnilingual;
     }
 
+    public OfflineMedAsrCtcModelConfig getMedAsr() {
+        return medasr;
+    }
+
     public OfflineCanaryModelConfig getCanary() {
         return canary;
     }
@@ -140,6 +146,7 @@ public class OfflineModelConfig {
         private OfflineZipformerCtcModelConfig zipformerCtc = OfflineZipformerCtcModelConfig.builder().build();
         private OfflineWenetCtcModelConfig wenetCtc = OfflineWenetCtcModelConfig.builder().build();
         private OfflineOmnilingualAsrCtcModelConfig omnilingual = OfflineOmnilingualAsrCtcModelConfig.builder().build();
+        private OfflineMedAsrCtcModelConfig medasr = OfflineMedAsrCtcModelConfig.builder().build();
         private OfflineCanaryModelConfig canary = OfflineCanaryModelConfig.builder().build();
         private String teleSpeech = "";
         private String tokens = "";
@@ -189,6 +196,11 @@ public class OfflineModelConfig {
             return this;
         }
 
+        public Builder setMedAsr(OfflineMedAsrCtcModelConfig medasr) {
+            this.medasr = medasr;
+            return this;
+        }
+
         public Builder setCanary(OfflineCanaryModelConfig canary) {
             this.canary = canary;
             return this;

sherpa-onnx/jni/offline-recognizer.cc

@@ -258,6 +258,15 @@ static OfflineRecognizerConfig GetOfflineConfig(JNIEnv *env, jobject config,
                               omnilingual_ctc_config_cls,
                               omnilingual_ctc_config);
 
+  // medasr ctc
+  fid = env->GetFieldID(model_config_cls, "medasr",
+                        "Lcom/k2fsa/sherpa/onnx/OfflineMedAsrCtcModelConfig;");
+  jobject medasr_ctc_config = env->GetObjectField(model_config, fid);
+  jclass medasr_ctc_config_cls = env->GetObjectClass(medasr_ctc_config);
+
+  SHERPA_ONNX_JNI_READ_STRING(ans.model_config.medasr.model, model,
+                              medasr_ctc_config_cls, medasr_ctc_config);
+
   // canary
   fid = env->GetFieldID(model_config_cls, "canary",
                         "Lcom/k2fsa/sherpa/onnx/OfflineCanaryModelConfig;");

sherpa-onnx/kotlin-api/OfflineRecognizer.kt

@@ -46,6 +46,10 @@ data class OfflineOmnilingualAsrCtcModelConfig(
     var model: String = "",
 )
 
+data class OfflineMedAsrCtcModelConfig(
+    var model: String = "",
+)
+
 data class OfflineWhisperModelConfig(
     var encoder: String = "",
     var decoder: String = "",
@@ -93,6 +97,7 @@ data class OfflineModelConfig(
     var zipformerCtc: OfflineZipformerCtcModelConfig = OfflineZipformerCtcModelConfig(),
     var wenetCtc: OfflineWenetCtcModelConfig = OfflineWenetCtcModelConfig(),
     var omnilingual: OfflineOmnilingualAsrCtcModelConfig = OfflineOmnilingualAsrCtcModelConfig(),
+    var medasr: OfflineMedAsrCtcModelConfig = OfflineMedAsrCtcModelConfig(),
     var canary: OfflineCanaryModelConfig = OfflineCanaryModelConfig(),
     var teleSpeech: String = "",
     var numThreads: Int = 1,
@@ -754,6 +759,16 @@ fun getOfflineModelConfig(type: Int): OfflineModelConfig? {
             )
         }
 
+        45 -> {
+            val modelDir = "sherpa-onnx-medasr-ctc-en-int8-2025-12-25"
+            return OfflineModelConfig(
+                medasr = OfflineMedAsrCtcModelConfig(
+                    model = "$modelDir/model.int8.onnx",
+                ),
+                tokens = "$modelDir/tokens.txt",
+            )
+        }
+
         9000 -> {
             val modelDir =
                 "sherpa-onnx-qnn-5-seconds-sense-voice-zh-en-ja-ko-yue-2024-07-17-int8-android-aarch64"