"""
本地TTS语音播报模块 - 基于IndexTTS2
"""
import os
import sys
import time
import tempfile
import winsound

# IndexTTS2 路径配置
_INDEXTTS_HOME = r"E:\2025Code\python\IndexTT\index-tts"
_INDEXTTS_MODEL_DIR = os.path.join(_INDEXTTS_HOME, "checkpoints")
_INDEXTTS_CFG = os.path.join(_INDEXTTS_MODEL_DIR, "config.yaml")
_INDEXTTS_VENV_PYTHON = os.path.join(_INDEXTTS_HOME, ".venv", "Scripts", "python.exe")

# 默认语音样本
_DEFAULT_VOICE_PROMPT = os.path.join(_INDEXTTS_HOME, "examples", "voice_01.wav")

# 全局 TTS 实例（延迟初始化）
_tts_instance = None


def _get_tts():
    """延迟初始化 IndexTTS2 实例"""
    global _tts_instance
    if _tts_instance is None:
        sys.path.insert(0, _INDEXTTS_HOME)
        from indextts.infer_v2 import IndexTTS2
        print("[TTS] 正在加载 IndexTTS2 模型（首次加载较慢，请耐心等待）...")
        _tts_instance = IndexTTS2(
            cfg_path=_INDEXTTS_CFG,
            model_dir=_INDEXTTS_MODEL_DIR,
            use_fp16=True,
            use_cuda_kernel=False,
            use_deepspeed=False
        )
        print("[TTS] IndexTTS2 模型加载完成")
    return _tts_instance


def speak(text, voice_prompt=None):
    """使用本地 IndexTTS2 合成语音并播放
    
    Args:
        text: 要播报的文本
        voice_prompt: 语音样本路径，默认使用 voice_01.wav
    """
    if voice_prompt is None:
        voice_prompt = _DEFAULT_VOICE_PROMPT
    
    tts = _get_tts()
    output_path = os.path.join(tempfile.gettempdir(), f"fx_tts_{int(time.time() * 1000)}.wav")
    
    tts.infer(
        spk_audio_prompt=voice_prompt,
        text=text,
        output_path=output_path,
        verbose=False
    )
    
    # 播放生成的音频
    winsound.PlaySound(output_path, winsound.SND_FILENAME)
    return output_path


if __name__ == "__main__":
    speak("本地语音播报测试成功")