Add Java API for text-to-speech (#811)

@@ -138,3 +138,21 @@ jobs:
 
           ./run-non-streaming-decode-file-nemo.sh
           rm -rf sherpa-onnx-nemo-*
+
+      - name:  Run java test (Non-Streaming TTS)
+        shell: bash
+        run: |
+          cd ./java-api-examples
+          ./run-non-streaming-tts-piper-en.sh
+          rm -rf vits-piper-*
+
+          ./run-non-streaming-tts-coqui-de.sh
+          rm -rf vits-coqui-*
+
+          ./run-non-streaming-tts-vits-zh.sh
+          rm -rf vits-zh-*
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: tts-wav-files-${{ matrix.os }}
+          path: java-api-examples/*.wav

+// Copyright 2024 Xiaomi Corporation
+
+// This file shows how to use a Coqui-ai VITS German TTS model
+// to convert text to speech
+import com.k2fsa.sherpa.onnx.*;
+
+public class NonStreamingTtsCoquiDe {
+  public static void main(String[] args) {
+    // please visit
+    // https://github.com/k2-fsa/sherpa-onnx/releases/tag/tts-models
+    // to download model files
+    String model = "./vits-coqui-de-css10/model.onnx";
+    String tokens = "./vits-coqui-de-css10/tokens.txt";
+    String text = "Alles hat ein Ende, nur die Wurst hat zwei.";
+
+    OfflineTtsVitsModelConfig vitsModelConfig =
+        OfflineTtsVitsModelConfig.builder().setModel(model).setTokens(tokens).build();
+
+    OfflineTtsModelConfig modelConfig =
+        OfflineTtsModelConfig.builder()
+            .setVits(vitsModelConfig)
+            .setNumThreads(1)
+            .setDebug(true)
+            .build();
+
+    OfflineTtsConfig config = OfflineTtsConfig.builder().setModel(modelConfig).build();
+    OfflineTts tts = new OfflineTts(config);
+
+    int sid = 0;
+    float speed = 1.0f;
+    long start = System.currentTimeMillis();
+    GeneratedAudio audio = tts.generate(text, sid, speed);
+    long stop = System.currentTimeMillis();
+
+    float timeElapsedSeconds = (stop - start) / 1000.0f;
+
+    float audioDuration = audio.getSamples().length / (float) audio.getSampleRate();
+    float real_time_factor = timeElapsedSeconds / audioDuration;
+
+    String waveFilename = "tts-coqui-de.wav";
+    audio.save(waveFilename);
+    System.out.printf("-- elapsed : %.3f seconds\n", timeElapsedSeconds);
+    System.out.printf("-- audio duration: %.3f seconds\n", timeElapsedSeconds);
+    System.out.printf("-- real-time factor (RTF): %.3f\n", real_time_factor);
+    System.out.printf("-- text: %s\n", text);
+    System.out.printf("-- Saved to %s\n", waveFilename);
+
+    tts.release();
+  }
+}

+// Copyright 2024 Xiaomi Corporation
+
+// This file shows how to use a piper VITS English TTS model
+// to convert text to speech
+import com.k2fsa.sherpa.onnx.*;
+
+public class NonStreamingTtsPiperEn {
+  public static void main(String[] args) {
+    // please visit
+    // https://github.com/k2-fsa/sherpa-onnx/releases/tag/tts-models
+    // to download model files
+    String model = "./vits-piper-en_GB-cori-medium/en_GB-cori-medium.onnx";
+    String tokens = "./vits-piper-en_GB-cori-medium/tokens.txt";
+    String dataDir = "./vits-piper-en_GB-cori-medium/espeak-ng-data";
+    String text =
+        "Today as always, men fall into two groups: slaves and free men. Whoever does not have"
+            + " two-thirds of his day for himself, is a slave, whatever he may be: a statesman, a"
+            + " businessman, an official, or a scholar.";
+
+    OfflineTtsVitsModelConfig vitsModelConfig =
+        OfflineTtsVitsModelConfig.builder()
+            .setModel(model)
+            .setTokens(tokens)
+            .setDataDir(dataDir)
+            .build();
+
+    OfflineTtsModelConfig modelConfig =
+        OfflineTtsModelConfig.builder()
+            .setVits(vitsModelConfig)
+            .setNumThreads(1)
+            .setDebug(true)
+            .build();
+
+    OfflineTtsConfig config = OfflineTtsConfig.builder().setModel(modelConfig).build();
+    OfflineTts tts = new OfflineTts(config);
+
+    int sid = 0;
+    float speed = 1.0f;
+    long start = System.currentTimeMillis();
+    GeneratedAudio audio = tts.generate(text, sid, speed);
+    long stop = System.currentTimeMillis();
+
+    float timeElapsedSeconds = (stop - start) / 1000.0f;
+
+    float audioDuration = audio.getSamples().length / (float) audio.getSampleRate();
+    float real_time_factor = timeElapsedSeconds / audioDuration;
+
+    String waveFilename = "tts-piper-en.wav";
+    audio.save(waveFilename);
+    System.out.printf("-- elapsed : %.3f seconds\n", timeElapsedSeconds);
+    System.out.printf("-- audio duration: %.3f seconds\n", timeElapsedSeconds);
+    System.out.printf("-- real-time factor (RTF): %.3f\n", real_time_factor);
+    System.out.printf("-- text: %s\n", text);
+    System.out.printf("-- Saved to %s\n", waveFilename);
+
+    tts.release();
+  }
+}

+// Copyright 2024 Xiaomi Corporation
+
+// This file shows how to use a VITS Chinese TTS model
+// to convert text to speech.
+//
+// You can use https://github.com/Plachtaa/VITS-fast-fine-tuning
+// to train your model
+import com.k2fsa.sherpa.onnx.*;
+
+public class NonStreamingTtsPiperEn {
+  public static void main(String[] args) {
+    // please visit
+    // https://github.com/k2-fsa/sherpa-onnx/releases/tag/tts-models
+    // to download model files
+    String model = "./vits-zh-hf-fanchen-C/vits-zh-hf-fanchen-C.onnx";
+    String tokens = "./vits-zh-hf-fanchen-C/tokens.txt";
+    String lexicon = "./vits-zh-hf-fanchen-C/lexicon.txt";
+    String dictDir = "./vits-zh-hf-fanchen-C/dict";
+    String ruleFsts =
+        "./vits-zh-hf-fanchen-C/phone.fst,./vits-zh-hf-fanchen-C/date.fst,./vits-zh-hf-fanchen-C/number.fst";
+    String text = "有问题，请拨打110或者手机18601239876。我们的价值观是真诚热爱！";
+
+    OfflineTtsVitsModelConfig vitsModelConfig =
+        OfflineTtsVitsModelConfig.builder()
+            .setModel(model)
+            .setTokens(tokens)
+            .setLexicon(lexicon)
+            .setDictDir(dictDir)
+            .build();
+
+    OfflineTtsModelConfig modelConfig =
+        OfflineTtsModelConfig.builder()
+            .setVits(vitsModelConfig)
+            .setNumThreads(1)
+            .setDebug(true)
+            .build();
+
+    OfflineTtsConfig config =
+        OfflineTtsConfig.builder().setModel(modelConfig).setRuleFsts(ruleFsts).build();
+
+    OfflineTts tts = new OfflineTts(config);
+
+    int sid = 100;
+    float speed = 1.0f;
+    long start = System.currentTimeMillis();
+    GeneratedAudio audio = tts.generate(text, sid, speed);
+    long stop = System.currentTimeMillis();
+
+    float timeElapsedSeconds = (stop - start) / 1000.0f;
+
+    float audioDuration = audio.getSamples().length / (float) audio.getSampleRate();
+    float real_time_factor = timeElapsedSeconds / audioDuration;
+
+    String waveFilename = "tts-vits-zh.wav";
+    audio.save(waveFilename);
+    System.out.printf("-- elapsed : %.3f seconds\n", timeElapsedSeconds);
+    System.out.printf("-- audio duration: %.3f seconds\n", timeElapsedSeconds);
+    System.out.printf("-- real-time factor (RTF): %.3f\n", real_time_factor);
+    System.out.printf("-- text: %s\n", text);
+    System.out.printf("-- Saved to %s\n", waveFilename);
+
+    tts.release();
+  }
+}

@@ -21,3 +21,11 @@ This directory contains examples for the JAVA API of sherpa-onnx.
 ./run-non-streaming-decode-file-whisper.sh
 ./run-non-streaming-decode-file-nemo.sh
 ```
+
+## Non-Streaming text-to-speech
+
+```bash
+./run-non-streaming-tts-piper-en.sh
+./run-non-streaming-tts-coqui-de.sh
+./run-non-streaming-tts-vits-zh.sh
+```

+#!/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 ../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
+if [ ! -f ./vits-coqui-de-css10/tokens.txt ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-coqui-de-css10.tar.bz2
+  tar xf vits-coqui-de-css10.tar.bz2
+  rm vits-coqui-de-css10.tar.bz2
+fi
+
+java \
+  -Djava.library.path=$PWD/../build/lib \
+  -cp ../sherpa-onnx/java-api/build/sherpa-onnx.jar \
+  NonStreamingTtsCoquiDe.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 ../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
+if [ ! -f ./vits-piper-en_GB-cori-medium/tokens.txt ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-piper-en_GB-cori-medium.tar.bz2
+  tar xf vits-piper-en_GB-cori-medium.tar.bz2
+  rm vits-piper-en_GB-cori-medium.tar.bz2
+fi
+
+java \
+  -Djava.library.path=$PWD/../build/lib \
+  -cp ../sherpa-onnx/java-api/build/sherpa-onnx.jar \
+  NonStreamingTtsPiperEn.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 ../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
+if [ ! -f ./vits-zh-hf-fanchen-C/tokens.txt ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/vits-zh-hf-fanchen-C.tar.bz2
+  tar xf vits-zh-hf-fanchen-C.tar.bz2
+  rm vits-zh-hf-fanchen-C.tar.bz2
+fi
+
+java \
+  -Djava.library.path=$PWD/../build/lib \
+  -cp ../sherpa-onnx/java-api/build/sherpa-onnx.jar \
+  NonStreamingTtsVitsZh.java

@@ -30,6 +30,12 @@ java_files += OfflineRecognizerResult.java
 java_files += OfflineStream.java
 java_files += OfflineRecognizer.java
 
+java_files += OfflineTtsVitsModelConfig.java
+java_files += OfflineTtsModelConfig.java
+java_files += OfflineTtsConfig.java
+java_files += GeneratedAudio.java
+java_files += OfflineTts.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 GeneratedAudio {
+    static {
+        System.loadLibrary("sherpa-onnx-jni");
+    }
+
+    private final float[] samples;
+    private final int sampleRate;
+
+    public GeneratedAudio(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);
+}

+// Copyright 2024 Xiaomi Corporation
+
+package com.k2fsa.sherpa.onnx;
+
+public class OfflineTts {
+    static {
+        System.loadLibrary("sherpa-onnx-jni");
+    }
+
+    private long ptr = 0; // this is the asr engine ptrss
+
+    public OfflineTts(OfflineTtsConfig config) {
+        ptr = newFromFile(config);
+    }
+
+    public GeneratedAudio generate(String text) {
+        return generate(text, 0, 1.0f);
+    }
+
+    public GeneratedAudio generate(String text, int sid) {
+        return generate(text, sid, 1.0f);
+    }
+
+    public GeneratedAudio generate(String text, int sid, float speed) {
+        Object[] arr = generateImpl(ptr, text, sid, speed);
+        float[] samples = (float[]) arr[0];
+        int sampleRate = (int) arr[1];
+        return new GeneratedAudio(samples, sampleRate);
+    }
+
+    @Override
+    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 int getNumSpeakers(long ptr);
+
+    private native Object[] generateImpl(long ptr, String text, int sid, float speed);
+
+    private native long newFromFile(OfflineTtsConfig config);
+}

+// Copyright 2024 Xiaomi Corporation
+
+package com.k2fsa.sherpa.onnx;
+
+public class OfflineTtsConfig {
+    private final OfflineTtsModelConfig model;
+    private final String ruleFsts;
+    private final String ruleFars;
+    private final int maxNumSentences;
+
+    private OfflineTtsConfig(Builder builder) {
+        this.model = builder.model;
+        this.ruleFsts = builder.ruleFsts;
+        this.ruleFars = builder.ruleFars;
+        this.maxNumSentences = builder.maxNumSentences;
+    }
+
+    public static Builder builder() {
+        return new Builder();
+    }
+
+    public OfflineTtsModelConfig getModel() {
+        return model;
+    }
+
+    public String getRuleFsts() {
+        return ruleFsts;
+    }
+
+    public String getRuleFars() {
+        return ruleFars;
+    }
+
+    public int getMaxNumSentences() {
+        return maxNumSentences;
+    }
+
+    public static class Builder {
+        private OfflineTtsModelConfig model = OfflineTtsModelConfig.builder().build();
+        private String ruleFsts = "";
+        private String ruleFars = "";
+        private int maxNumSentences = 1;
+
+        public OfflineTtsConfig build() {
+            return new OfflineTtsConfig(this);
+        }
+
+        public Builder setModel(OfflineTtsModelConfig model) {
+            this.model = model;
+            return this;
+        }
+
+        public Builder setRuleFsts(String ruleFsts) {
+            this.ruleFsts = ruleFsts;
+            return this;
+        }
+
+        public Builder setRuleFars(String ruleFars) {
+            this.ruleFars = ruleFars;
+            return this;
+        }
+
+        public Builder setMaxNumSentences(int maxNumSentences) {
+            this.maxNumSentences = maxNumSentences;
+            return this;
+        }
+    }
+}

+// Copyright 2024 Xiaomi Corporation
+
+package com.k2fsa.sherpa.onnx;
+
+public class OfflineTtsModelConfig {
+    private final OfflineTtsVitsModelConfig vits;
+    private final int numThreads;
+    private final boolean debug;
+    private final String provider;
+
+    private OfflineTtsModelConfig(Builder builder) {
+        this.vits = builder.vits;
+        this.numThreads = builder.numThreads;
+        this.debug = builder.debug;
+        this.provider = builder.provider;
+    }
+
+    public static Builder builder() {
+        return new Builder();
+    }
+
+    public OfflineTtsVitsModelConfig getVits() {
+        return vits;
+    }
+
+    public static class Builder {
+        private OfflineTtsVitsModelConfig vits = OfflineTtsVitsModelConfig.builder().build();
+        private int numThreads = 1;
+        private boolean debug = true;
+        private String provider = "cpu";
+
+        public OfflineTtsModelConfig build() {
+            return new OfflineTtsModelConfig(this);
+        }
+
+        public Builder setVits(OfflineTtsVitsModelConfig vits) {
+            this.vits = vits;
+            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 2024 Xiaomi Corporation
+
+package com.k2fsa.sherpa.onnx;
+
+public class OfflineTtsVitsModelConfig {
+    private final String model;
+    private final String lexicon;
+    private final String tokens;
+    private final String dataDir;
+    private final String dictDir;
+    private final float noiseScale;
+    private final float noiseScaleW;
+    private final float lengthScale;
+
+    private OfflineTtsVitsModelConfig(Builder builder) {
+        this.model = builder.model;
+        this.lexicon = builder.lexicon;
+        this.tokens = builder.tokens;
+        this.dataDir = builder.dataDir;
+        this.dictDir = builder.dictDir;
+        this.noiseScale = builder.noiseScale;
+        this.noiseScaleW = builder.noiseScaleW;
+        this.lengthScale = builder.lengthScale;
+    }
+
+    public static Builder builder() {
+        return new Builder();
+    }
+
+    public String getModel() {
+        return model;
+    }
+
+    public String getLexicon() {
+        return lexicon;
+    }
+
+    public String getTokens() {
+        return tokens;
+    }
+
+    public String getDataDir() {
+        return dataDir;
+    }
+
+    public String getDictDir() {
+        return dictDir;
+    }
+
+    public float getLengthScale() {
+        return lengthScale;
+    }
+
+    public float getNoiseScale() {
+        return noiseScale;
+    }
+
+    public float getNoiseScaleW() {
+        return noiseScaleW;
+    }
+
+    public static class Builder {
+        private String model;
+        private String lexicon = "";
+        private String tokens;
+        private String dataDir = "";
+        private String dictDir = "";
+        private float noiseScale = 0.667f;
+        private float noiseScaleW = 0.8f;
+        private float lengthScale = 1.0f;
+
+        public OfflineTtsVitsModelConfig build() {
+            return new OfflineTtsVitsModelConfig(this);
+        }
+
+        public Builder setModel(String model) {
+            this.model = model;
+            return this;
+        }
+
+        public Builder setTokens(String tokens) {
+            this.tokens = tokens;
+            return this;
+        }
+
+        public Builder setLexicon(String lexicon) {
+            this.lexicon = lexicon;
+            return this;
+        }
+
+        public Builder setDataDir(String dataDir) {
+            this.dataDir = dataDir;
+            return this;
+        }
+
+        public Builder setDictDir(String dictDir) {
+            this.dictDir = dictDir;
+            return this;
+        }
+
+        public Builder setNoiseScale(float noiseScale) {
+            this.noiseScale = noiseScale;
+            return this;
+        }
+
+        public Builder setNoiseScaleW(float noiseScaleW) {
+            this.noiseScaleW = noiseScaleW;
+            return this;
+        }
+
+        public Builder setLengthScale(float lengthScale) {
+            this.lengthScale = lengthScale;
+            return this;
+        }
+    }
+}

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