app.py

import gradio as gr
from TTS.api import TTS
import numpy as np
import os

# 设置TTS_HOME并创建tos_agreed文件以绕过许可提示
tts_home = os.path.expanduser("~/.tts")
os.makedirs(tts_home, exist_ok=True)
tos_agreed_file = os.path.join(tts_home, "tos_agreed")
if not os.path.exists(tos_agreed_file):
    with open(tos_agreed_file, "w") as f:
        f.write("I agree to the terms of the non-commercial CPML: https://coqui.ai/cpml")

# 初始化TTS模型，使用支持中文的XTTS模型
tts = TTS(model_name="tts_models/multilingual/multi-dataset/xtts_v2", progress_bar=False, gpu=False)

def text_to_speech(text):
    try:
        # 确保输入文本是字符串
        if not isinstance(text, str):
            return None, "请输入有效的中文文本"
        
        # 生成语音，指定语言为zh-cn
        wav = tts.tts(text=text, language="zh-cn")
        
        # 保存为wav文件
        output_path = "output.wav"
        tts.save_wav(wav, output_path)
        
        return output_path, "语音生成成功"
    except Exception as e:
        return None, f"生成失败，可能由于文本内容不合适或模型限制: {str(e)}"

# Gradio界面
with gr.Blocks() as demo:
    gr.Markdown("# 中文语音生成器 _

    gr.Markdown("输入中文文本，生成对应的语音。")
    
    with gr.Row():
        text_input = gr.Textbox(label="输入中文文本", placeholder="请输入要转换的中文文本...")
        submit_button = gr.Button("生成语音")
    
    with gr.Row():
        audio_output = gr.Audio(label="生成的语音")
        status_output = gr.Textbox(label="状态")
    
    submit_button.click(
        fn=text_to_speech,
        inputs=text_input,
        outputs=[audio_output, status_output]
    )

# 启动Gradio应用
demo.launch()