Add Python API examples for Supertonic TTS

Author: csukuangfj

.github/scripts/test-python.sh

@@ -8,6 +8,20 @@ log() {
   echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
 }
 
+log "test Supertonic TTS"
+
+curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/sherpa-onnx-supertonic-tts-int8-2026-03-06.tar.bz2
+tar xvf sherpa-onnx-supertonic-tts-int8-2026-03-06.tar.bz2
+rm sherpa-onnx-supertonic-tts-int8-2026-03-06.tar.bz2
+
+python3 ./supertonic-tts.py
+
+rm -rf sherpa-onnx-supertonic-tts-int8-2026-03-06
+
+mkdir -p tts
+cp supertonic-en.wav tts/
+ls -lh tts
+
 log "test Moonshine v2"
 
 curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-moonshine-tiny-en-quantized-2026-02-27.tar.bz2
@@ -399,7 +413,7 @@ done
 
 log "Offline TTS test"
 # test waves are saved in ./tts
-mkdir ./tts
+mkdir -p ./tts
 
 log "test kitten tts"
 

.gitignore

@@ -176,3 +176,4 @@ sherpa-onnx-sense-voice-zh-en-ja-ko-yue-int8-2024-07-17
 sherpa-onnx-fire-red-asr2-ctc-zh_en-int8-2026-02-25
 non-streaming-fire-red-asr-ctc-decode-files
 sherpa-onnx-moonshine-*-quantized-2026-02-27
+sherpa-onnx-supertonic-tts-int8-2026-03-06

python-api-examples/supertonic-tts.py

@@ -0,0 +1,102 @@
+#!/usr/bin/env python3
+#
+# Copyright (c)  2026  Xiaomi Corporation
+
+"""
+This file demonstrates how to use sherpa-onnx Python API
+for SupertonicTTS.
+
+
+Usage:
+
+wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/sherpa-onnx-supertonic-tts-int8-2026-03-06.tar.bz2
+tar xvf sherpa-onnx-supertonic-tts-int8-2026-03-06.tar.bz2
+rm sherpa-onnx-supertonic-tts-int8-2026-03-06.tar.bz2
+
+python3 ./supertonic-tts.py
+
+You can find more models at
+https://github.com/k2-fsa/sherpa-onnx/releases/tag/tts-models
+
+Please see
+https://k2-fsa.github.io/sherpa/onnx/tts/supertonic.html
+for details.
+
+"""
+
+import time
+
+import sherpa_onnx
+import soundfile as sf
+
+
+def create_tts():
+    tts_config = sherpa_onnx.OfflineTtsConfig(
+        model=sherpa_onnx.OfflineTtsModelConfig(
+            supertonic=sherpa_onnx.OfflineTtsSupertonicModelConfig(
+                duration_predictor="./sherpa-onnx-supertonic-tts-int8-2026-03-06/duration_predictor.int8.onnx",
+                text_encoder="./sherpa-onnx-supertonic-tts-int8-2026-03-06/text_encoder.int8.onnx",
+                vector_estimator="./sherpa-onnx-supertonic-tts-int8-2026-03-06/vector_estimator.int8.onnx",
+                vocoder="./sherpa-onnx-supertonic-tts-int8-2026-03-06/vocoder.int8.onnx",
+                tts_json="./sherpa-onnx-supertonic-tts-int8-2026-03-06/tts.json",
+                unicode_indexer="./sherpa-onnx-supertonic-tts-int8-2026-03-06/unicode_indexer.bin",
+                voice_style="./sherpa-onnx-supertonic-tts-int8-2026-03-06/voice.bin",
+            ),
+            debug=False,
+            num_threads=2,
+            provider="cpu",
+        )
+    )
+    if not tts_config.validate():
+        raise ValueError(
+            "Please read the previous error messages and re-check your config"
+        )
+
+    return sherpa_onnx.OfflineTts(tts_config)
+
+
+def main():
+    tts = create_tts()
+
+    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."
+
+    gen_config = sherpa_onnx.GenerationConfig()
+
+    # This model has 10 speakers. Valid sid: 0-9
+    gen_config.sid = 6
+    gen_config.num_steps = 5
+    gen_config.speed = 1.25  # larger -> faster
+
+    # We use en for English.
+    # You can also use es, pt, fr, ko.
+    # This single model supports 5 languages.
+    gen_config.extra["lang"] = "en"
+
+    start = time.time()
+    audio = tts.generate(text, gen_config)
+    end = time.time()
+
+    if len(audio.samples) == 0:
+        print("Error in generating audios. Please read previous error messages.")
+        return
+
+    elapsed_seconds = end - start
+    audio_duration = len(audio.samples) / audio.sample_rate
+    real_time_factor = elapsed_seconds / audio_duration
+
+    output_filename = "./supertonic-en.wav"
+    sf.write(
+        output_filename,
+        audio.samples,
+        samplerate=audio.sample_rate,
+        subtype="PCM_16",
+    )
+    print(f"Saved to {output_filename}")
+    print(f"The text is '{text}'")
+    print(f"Elapsed seconds: {elapsed_seconds:.3f}")
+    print(f"Audio duration in seconds: {audio_duration:.3f}")
+    print(f"RTF: {elapsed_seconds:.3f}/{audio_duration:.3f} = {real_time_factor:.3f}")
+
+
+if __name__ == "__main__":
+    main()

sherpa-onnx/csrc/offline-tts-supertonic-impl.cc

@@ -603,6 +603,10 @@ void OfflineTtsSupertonicImpl::InitVoiceStyle(const std::vector<char> &buf) {
   }
   num_speakers_ = num_speakers;
   full_style_ = std::move(style);
+
+  if (config_.model.debug) {
+    SHERPA_ONNX_LOGE("Number of speakers: %d", num_speakers_);
+  }
 }
 
 OfflineTtsSupertonicImpl::StyleSliceView

sherpa-onnx/python/sherpa_onnx/__init__.py

@@ -51,6 +51,7 @@
     OfflineTtsMatchaModelConfig,
     OfflineTtsModelConfig,
     OfflineTtsPocketModelConfig,
+    OfflineTtsSupertonicModelConfig,
     OfflineTtsVitsModelConfig,
     OfflineTtsZipvoiceModelConfig,
     OfflineWenetCtcModelConfig,