sherpa-onnx/java-api-examples/NonStreamingTtsKokoroZhEnLexicon.java

// Copyright 2025 Xiaomi Corporation

// This file shows how to use a Kokoro multi-lingual model
// to convert Chinese and English text to speech
import com.k2fsa.sherpa.onnx.*;

public class NonStreamingTtsKokoroZhEnLexicon {
  public static void main(String[] args) {
    // please visit
    // https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/kokoro.html
    // to download model files
    String model = "./kokoro-multi-lang-v1_0/model.onnx";
    String voices = "./kokoro-multi-lang-v1_0/voices.bin";
    String tokens = "./kokoro-multi-lang-v1_0/tokens.txt";
    String lexicon =
        "./kokoro-multi-lang-v1_0/lexicon-us-en.txt,./kokoro-multi-lang-v1_0/lexicon-zh.txt";
    String text =
        "中英文语音合成测试。This is generated by next generation Kaldi using Kokoro without Misaki."
            + " 你觉得中英文说的如何呢？";

    OfflineTtsKokoroModelConfig kokoroModelConfig =
        OfflineTtsKokoroModelConfig.builder()
            .setModel(model)
            .setVoices(voices)
            .setTokens(tokens)
            .setLexicon(lexicon)
            .build();

    OfflineTtsModelConfig modelConfig =
        OfflineTtsModelConfig.builder()
            .setKokoro(kokoroModelConfig)
            .setNumThreads(2)
            .setDebug(true)
            .build();

    OfflineTtsConfig config = OfflineTtsConfig.builder().setModel(modelConfig).build();
    OfflineTts tts = new OfflineTts(config);

    int sid = 0; // this model has 53 speakers. You can use sid in the range 0-52
    float speed = 1.0f;
    GenerationConfig genConfig = new GenerationConfig();
    genConfig.setSid(sid);
    genConfig.setSpeed(speed);
    genConfig.setSilenceScale(0.2f);
    long start = System.currentTimeMillis();
    GeneratedAudio audio = tts.generateWithConfigAndCallback(text, genConfig, samples -> {});
    long stop = System.currentTimeMillis();

    float timeElapsedSeconds = (stop - start) / 1000.0f;

    float audioDuration = audio.getSamples().length / (float) audio.getSampleRate();
    float realTimeFactor = timeElapsedSeconds / audioDuration;

    String waveFilename = "tts-kokoro-zh-en-lexicon.wav";
    audio.save(waveFilename);
    System.out.printf("-- elapsed : %.3f seconds\n", timeElapsedSeconds);
    System.out.printf("-- audio duration: %.3f seconds\n", audioDuration);
    System.out.printf("-- real-time factor (RTF): %.3f\n", realTimeFactor);
    System.out.printf("-- text: %s\n", text);
    System.out.printf("-- Saved to %s\n", waveFilename);

    tts.release();
  }
}