Add Java and Kotlin API for PocketTTS (#3095)

java-api-examples/PocketTts.java

@@ -0,0 +1,147 @@
+// Copyright 2026 Xiaomi Corporation
+
+// This file shows how to use a PocketTTS English model
+// for voice cloning.
+import com.k2fsa.sherpa.onnx.*;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.function.Consumer;
+import java.util.function.Function;
+
+public class PocketTts {
+  public static void main(String[] args) {
+    LibraryUtils.enableDebug();
+    // please visit
+    // https://k2-fsa.github.io/sherpa/onnx/tts/pocket.html
+    // to download model files
+    String lmFlow = "./sherpa-onnx-pocket-tts-int8-2026-01-26/lm_flow.int8.onnx";
+    String lmMain = "./sherpa-onnx-pocket-tts-int8-2026-01-26/lm_main.int8.onnx";
+    String encoder = "./sherpa-onnx-pocket-tts-int8-2026-01-26/encoder.onnx";
+    String decoder = "./sherpa-onnx-pocket-tts-int8-2026-01-26/decoder.int8.onnx";
+    String textConditioner = "./sherpa-onnx-pocket-tts-int8-2026-01-26/text_conditioner.onnx";
+    String vocabJson = "./sherpa-onnx-pocket-tts-int8-2026-01-26/vocab.json";
+    String tokenScoresJson = "./sherpa-onnx-pocket-tts-int8-2026-01-26/token_scores.json";
+    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.";
+
+    OfflineTtsPocketModelConfig pocketModelConfig =
+        OfflineTtsPocketModelConfig.builder()
+            .setLmMain(lmMain)
+            .setLmFlow(lmFlow)
+            .setEncoder(encoder)
+            .setDecoder(decoder)
+            .setTextConditioner(textConditioner)
+            .setVocabJson(vocabJson)
+            .setTokenScoresJson(tokenScoresJson)
+            .build();
+
+    OfflineTtsModelConfig modelConfig =
+        OfflineTtsModelConfig.builder()
+            .setPocket(pocketModelConfig)
+            .setNumThreads(2)
+            .setDebug(true)
+            .build();
+
+    OfflineTtsConfig config = OfflineTtsConfig.builder().setModel(modelConfig).build();
+    OfflineTts tts = new OfflineTts(config);
+
+    String referenceAudioFilename = "./sherpa-onnx-pocket-tts-int8-2026-01-26/test_wavs/bria.wav";
+    WaveReader reader = new WaveReader(referenceAudioFilename);
+
+    GenerationConfig genConfig = new GenerationConfig();
+    genConfig.setReferenceAudio(reader.getSamples());
+    genConfig.setReferenceSampleRate(reader.getSampleRate());
+    genConfig.setNumSteps(5);
+
+    Map<String, String> extra = new HashMap<>();
+    extra.put("temperature", "0.7");
+    extra.put("chunk_size", "15");
+
+    genConfig.setExtra(extra);
+
+    long start = System.currentTimeMillis();
+    GeneratedAudio audio = null;
+
+    // You can choose one of the following callback style
+    // ---------------------------------------------------
+    // 1. Anonymous class implementing OfflineTtsCallback
+    // ---------------------------------------------------
+    if (true) {
+      audio =
+          tts.generateWithConfigAndCallback(
+              text,
+              genConfig,
+              new OfflineTtsCallback() {
+                @Override
+                public Integer invoke(float[] samples) {
+                  // you can play the generated samples in a separate thread
+                  System.out.println("callback got called with " + samples.length + " samples");
+                  // 1 = continue, 0 = stop
+                  return 1;
+                }
+              });
+    }
+
+    // -------------------------------
+    // 2. Lambda implementing OfflineTtsCallback
+    // -------------------------------
+    if (false) {
+      audio =
+          tts.generateWithConfigAndCallback(
+              text,
+              genConfig,
+              (OfflineTtsCallback)
+                  samples -> {
+                    System.out.println("Lambda Integer callback: " + samples.length);
+                    return 1; // continue
+                  });
+    }
+
+    if (false) {
+      audio =
+          tts.generateWithConfigAndCallback(
+              text,
+              genConfig,
+              (Consumer<float[]>)
+                  samples -> {
+                    System.out.println("Consumer: " + samples.length);
+                  });
+    }
+
+    if (false) {
+      audio =
+          tts.generateWithConfigAndCallback(
+              text,
+              genConfig,
+              (Function<float[], Integer>)
+                  samples -> {
+                    System.out.println("Function: " + samples.length);
+                    return 1;
+                  });
+    }
+
+    if (audio == null) {
+      System.err.println("No audio was generated. Please enable at least one callback branch.");
+      return;
+    }
+
+    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 = "pocket-tts-bria.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", real_time_factor);
+    System.out.printf("-- text: %s\n", text);
+    System.out.printf("-- Saved to %s\n", waveFilename);
+
+    tts.release();
+  }
+}