Add Java API for speech enhancement GTCRN models (#2009)

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

@@ -105,6 +105,15 @@ jobs:
             make -j4
             ls -lh lib
 
+      - name:  Run speech enhancement (GTCRN)
+        shell: bash
+        run: |
+          cd ./java-api-examples
+          ./run-non-streaming-speech-enhancement-gtcrn.sh
+          ls -lh *.wav
+
+          rm -fv gtcrn_simple.onnx *.wav
+
       - name:  Run java test (Online add punctuations)
         shell: bash
         run: |

java-api-examples/NonStreamingSpeechEnhancementGtcrn.java

@@ -0,0 +1,38 @@
+// Copyright 2025 Xiaomi Corporation
+
+// This file shows how to use speech enhancement models in sherpa-onnx
+//
+// please download files in this script from
+// https://github.com/k2-fsa/sherpa-onnx/releases/tag/speech-enhancement-models
+
+import com.k2fsa.sherpa.onnx.*;
+
+public class NonStreamingSpeechEnhancementGtcrn {
+  public static void main(String[] args) {
+    String model = "./gtcrn_simple.onnx";
+    OfflineSpeechDenoiserGtcrnModelConfig gtcrn =
+        OfflineSpeechDenoiserGtcrnModelConfig.builder().setModel(model).build();
+
+    OfflineSpeechDenoiserModelConfig modelConfig =
+        OfflineSpeechDenoiserModelConfig.builder()
+            .setGtcrn(gtcrn)
+            .setNumThreads(1)
+            .setDebug(true)
+            .setProvider("cpu")
+            .build();
+    OfflineSpeechDenoiserConfig config =
+        OfflineSpeechDenoiserConfig.builder().setModel(modelConfig).build();
+
+    OfflineSpeechDenoiser speech_denoiser = new OfflineSpeechDenoiser(config);
+
+    String testWaveFilename = "./inp_16k.wav";
+    WaveReader reader = new WaveReader(testWaveFilename);
+
+    DenoisedAudio denoised = speech_denoiser.run(reader.getSamples(), reader.getSampleRate());
+    String outFilename = "enhanced-16k.wav";
+    WaveWriter.write(outFilename, denoised.getSamples(), denoised.getSampleRate());
+    System.out.printf("Saved to %s\n", outFilename);
+
+    speech_denoiser.release();
+  }
+}

java-api-examples/run-non-streaming-speech-enhancement-gtcrn.sh

@@ -0,0 +1,39 @@
+#!/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 ./gtcrn_simple.onnx ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/speech-enhancement-models/gtcrn_simple.onnx
+fi
+
+if [ ! -f ./inp_16k.wav ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/speech-enhancement-models/inp_16k.wav
+fi
+
+java \
+  -Djava.library.path=$PWD/../build/lib \
+  -cp ../sherpa-onnx/java-api/build/sherpa-onnx.jar \
+  NonStreamingSpeechEnhancementGtcrn.java

sherpa-onnx/csrc/offline-speech-denoiser-gtcrn-model.cc

@@ -9,6 +9,15 @@
 #include <utility>
 #include <vector>
 
+#if __ANDROID_API__ >= 9
+#include "android/asset_manager.h"
+#include "android/asset_manager_jni.h"
+#endif
+
+#if __OHOS__
+#include "rawfile/raw_file_manager.h"
+#endif
+
 #include "sherpa-onnx/csrc/file-utils.h"
 #include "sherpa-onnx/csrc/onnx-utils.h"
 #include "sherpa-onnx/csrc/session.h"
@@ -193,4 +202,14 @@ OfflineSpeechDenoiserGtcrnModel::GetMetaData() const {
   return impl_->GetMetaData();
 }
 
+#if __ANDROID_API__ >= 9
+template OfflineSpeechDenoiserGtcrnModel::OfflineSpeechDenoiserGtcrnModel(
+    AAssetManager *mgr, const OfflineSpeechDenoiserModelConfig &config);
+#endif
+
+#if __OHOS__
+template OfflineSpeechDenoiserGtcrnModel::OfflineSpeechDenoiserGtcrnModel(
+    NativeResourceManager *mgr, const OfflineSpeechDenoiserModelConfig &config);
+#endif
+
 }  // namespace sherpa_onnx

sherpa-onnx/java-api/Makefile

@@ -84,6 +84,11 @@ java_files += OfflineSpeakerDiarizationSegment.java
 java_files += OfflineSpeakerDiarizationCallback.java
 java_files += OfflineSpeakerDiarization.java
 
+java_files += OfflineSpeechDenoiserGtcrnModelConfig.java
+java_files += OfflineSpeechDenoiserModelConfig.java
+java_files += OfflineSpeechDenoiserConfig.java
+java_files += DenoisedAudio.java
+java_files += OfflineSpeechDenoiser.java
 
 class_files := $(java_files:%.java=%.class)
 

sherpa-onnx/java-api/src/com/k2fsa/sherpa/onnx/DenoisedAudio.java

@@ -0,0 +1,32 @@
+// Copyright 2025 Xiaomi Corporation
+
+package com.k2fsa.sherpa.onnx;
+
+public class DenoisedAudio {
+    static {
+        System.loadLibrary("sherpa-onnx-jni");
+    }
+
+    private final float[] samples;
+    private final int sampleRate;
+
+    public DenoisedAudio(float[] samples, int sampleRate) {
+        this.samples = samples;
+        this.sampleRate = sampleRate;
+    }
+
+    public int getSampleRate() {
+        return sampleRate;
+    }
+
+    public float[] getSamples() {
+        return samples;
+    }
+
+    // return true if saved successfully.
+    public boolean save(String filename) {
+        return saveImpl(filename, samples, sampleRate);
+    }
+
+    private native boolean saveImpl(String filename, float[] samples, int sampleRate);
+}

sherpa-onnx/java-api/src/com/k2fsa/sherpa/onnx/OfflineSpeechDenoiser.java

@@ -0,0 +1,43 @@
+// Copyright 2025 Xiaomi Corporation
+
+package com.k2fsa.sherpa.onnx;
+
+public class OfflineSpeechDenoiser {
+    static {
+        System.loadLibrary("sherpa-onnx-jni");
+    }
+
+    private long ptr = 0;
+
+    public OfflineSpeechDenoiser(OfflineSpeechDenoiserConfig config) {
+        ptr = newFromFile(config);
+    }
+
+    public int getSampleRate() {
+        return getSampleRate(ptr);
+    }
+
+    public DenoisedAudio run(float[] samples, int sampleRate) {
+        return run(ptr, samples, sampleRate);
+    }
+
+    protected void finalize() throws Throwable {
+        release();
+    }
+
+    public void release() {
+        if (this.ptr == 0) {
+            return;
+        }
+        delete(this.ptr);
+        this.ptr = 0;
+    }
+
+    private native void delete(long ptr);
+
+    private native int getSampleRate(long ptr);
+
+    private native DenoisedAudio run(long ptr, float[] samples, int sampleRate);
+
+    private native long newFromFile(OfflineSpeechDenoiserConfig config);
+}

sherpa-onnx/java-api/src/com/k2fsa/sherpa/onnx/OfflineSpeechDenoiserConfig.java

@@ -0,0 +1,27 @@
+// Copyright 2025 Xiaomi Corporation
+package com.k2fsa.sherpa.onnx;
+
+public class OfflineSpeechDenoiserConfig {
+    private final OfflineSpeechDenoiserModelConfig model;
+
+    private OfflineSpeechDenoiserConfig(OfflineSpeechDenoiserConfig.Builder builder) {
+        this.model = builder.model;
+    }
+
+    public static Builder builder() {
+        return new Builder();
+    }
+
+    public static class Builder {
+        private OfflineSpeechDenoiserModelConfig model = OfflineSpeechDenoiserModelConfig.builder().build();
+
+        public OfflineSpeechDenoiserConfig build() {
+            return new OfflineSpeechDenoiserConfig(this);
+        }
+
+        public Builder setModel(OfflineSpeechDenoiserModelConfig model) {
+            this.model = model;
+            return this;
+        }
+    }
+}

sherpa-onnx/java-api/src/com/k2fsa/sherpa/onnx/OfflineSpeechDenoiserGtcrnModelConfig.java

@@ -0,0 +1,31 @@
+// Copyright 2025 Xiaomi Corporation
+package com.k2fsa.sherpa.onnx;
+
+public class OfflineSpeechDenoiserGtcrnModelConfig {
+    private final String model;
+
+    private OfflineSpeechDenoiserGtcrnModelConfig(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 OfflineSpeechDenoiserGtcrnModelConfig build() {
+            return new OfflineSpeechDenoiserGtcrnModelConfig(this);
+        }
+
+        public Builder setModel(String model) {
+            this.model = model;
+            return this;
+        }
+    }
+}

sherpa-onnx/java-api/src/com/k2fsa/sherpa/onnx/OfflineSpeechDenoiserModelConfig.java

@@ -0,0 +1,51 @@
+// Copyright 2025 Xiaomi Corporation
+package com.k2fsa.sherpa.onnx;
+
+public class OfflineSpeechDenoiserModelConfig {
+    private final OfflineSpeechDenoiserGtcrnModelConfig gtcrn;
+    private final int numThreads;
+    private final boolean debug;
+    private final String provider;
+
+    private OfflineSpeechDenoiserModelConfig(Builder builder) {
+        this.gtcrn = builder.gtcrn;
+        this.numThreads = builder.numThreads;
+        this.debug = builder.debug;
+        this.provider = builder.provider;
+    }
+
+    public static Builder builder() {
+        return new Builder();
+    }
+
+    public static class Builder {
+        private OfflineSpeechDenoiserGtcrnModelConfig gtcrn = OfflineSpeechDenoiserGtcrnModelConfig.builder().build();
+        private int numThreads = 1;
+        private boolean debug = true;
+        private String provider = "cpu";
+
+        public OfflineSpeechDenoiserModelConfig build() {
+            return new OfflineSpeechDenoiserModelConfig(this);
+        }
+
+        public Builder setGtcrn(OfflineSpeechDenoiserGtcrnModelConfig gtcrn) {
+            this.gtcrn = gtcrn;
+            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;
+        }
+    }
+}