uv init qwen3-tts
uv venv -p 3.12
uv pip install torch torchvision torchaudio ffmpeg qwen-tts

# ==========================================
# qwen3-tts voice clone program
# PLEASE change variable on "1. Set Var"
# ==========================================
import torch
import soundfile as sf
from qwen_tts import Qwen3TTSModel

# ==========================================
# 1. Set Variables
# ==========================================
ref_audio_path = "./sample_sound.wav"
ref_text = "こんにちは、かめりあんです。一般成人男性です。"
target_text = "ドーナツみたいな雲は、僕の心のようだった"
output_file="output_voiceClone.wav"

# ==========================================
# 2. Load Qwen3-TTS
# ==========================================
model = Qwen3TTSModel.from_pretrained(
  "Qwen/Qwen3-TTS-12Hz-1.7B-Base",
  device_map="cuda:0",
  dtype=torch.bfloat16,
)

# ==========================================
# 3. Run voice clone
# ==========================================
wavs, sr = model.generate_voice_clone(
  text=target_text,
  language="Japanese",
  ref_audio=ref_audio_path,
  ref_text=ref_text,
)

# ==========================================
# 4. save output wave file
# ==========================================
sf.write(output_file, wavs[0], sr)
print(f"output has been generated in ./{output_file}")