app.py

import gradio as gr
from TTS.utils.manage import ModelManager

@staticmethod
def new_ask_tos(model_full_path):
    return True

ModelManager.ask_tos = new_ask_tos

from TTS.api import TTS
import torch
import os
import random
import string
from cn_tx import TextNorm
normalizer = TextNorm()

def preprocess(text):
    # text = text.replace("掌柜", "涨柜")
    text = normalizer(text)
    return text
    
def tts_function(voice_file, text):
    speaker_wav = voice_file
    text = preprocess(text)
    # Generate a random filename with 16 characters
    random_filename = ''.join(random.choices(
        string.ascii_letters + string.digits, k=16))
    file_path = f"output/{random_filename}.wav"
    if not os.path.exists("output"):
        os.makedirs("output")

    # Run TTS
    tts.tts_to_file(text=text, speaker_wav=speaker_wav,
                    language="zh-cn", file_path=file_path)

    return file_path


# Create the Gradio interface with Chinese labels and placeholders
iface = gr.Interface(
    fn=tts_function,
    inputs=[
        gr.Audio(sources=["upload", "microphone"],
                 type="filepath", label="上传或录制声音"),
        gr.Textbox(lines=2, placeholder="请输入要转换的文字...", label="文本输入"),
    ],
    outputs=gr.Audio(type="filepath", label="生成的语音"),
    title="数字栩生声音克隆",
    allow_flagging='never'
)

if __name__ == '__main__':
    # Initialize TTS
    device = "cuda" if torch.cuda.is_available() else "cpu"
    tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)
    # tts = None
    iface.launch(inbrowser=True)