app.py

import gradio as gr
import numpy as np
import torch
import os
import re_matching
from tools.sentence import split_by_language, sentence_split
import utils
from infer import infer, latest_version, get_net_g
import gradio as gr
import webbrowser
from config import config
from tools.translate import translate

from webui import reload_javascript

device = config.webui_config.device
if device == "mps":
    os.environ["PYTORCH_ENABLE_MPS_FALLBACK"] = "1"

def generate_audio(
    slices,
    sdp_ratio,
    noise_scale,
    noise_scale_w,
    length_scale,
    speaker,
    language,
):
    audio_list = []
    silence = np.zeros(hps.data.sampling_rate // 2, dtype=np.int16)
    with torch.no_grad():
        for piece in slices:
            audio = infer(
                piece,
                sdp_ratio=sdp_ratio,
                noise_scale=noise_scale,
                noise_scale_w=noise_scale_w,
                length_scale=length_scale,
                sid=speaker,
                language=language,
                hps=hps,
                net_g=net_g,
                device=device,
            )
            audio16bit = gr.processing_utils.convert_to_16_bit_wav(audio)
            audio_list.append(audio16bit)
            audio_list.append(silence)  # 将静音添加到列表中
    return audio_list

def speak(
        text: str,
        speaker="TalkFlower_CNzh",
        sdp_ratio=0.2,      # SDP/DP混合比
        noise_scale=0.6,        # 感情
        noise_scale_w=0.6,      # 音素长度
        length_scale=0.9,       # 语速
        language="ZH"
    ):
    audio_list = []
    audio_list.extend(
        generate_audio(
            text.split("|"),
            sdp_ratio,
            noise_scale,
            noise_scale_w,
            length_scale,
            speaker,
            language,
        )
    )
    
    audio_concat = np.concatenate(audio_list)    
    return (hps.data.sampling_rate, audio_concat)


with open("./css/style.css", "r", encoding="utf-8") as f:
    customCSS = f.read()

with gr.Blocks(css=customCSS) as demo:
    # talkingFlowerModel = gr.HTML("""<div id="talking_flower_model">123</div>""")
    talkingFlowerPic = gr.HTML("""<img src="file=assets/flower-2x.webp" alt="TalkingFlowerPic">""", elem_id="talking_flower_pic")
    input_text = gr.Textbox(lines=1, label="Talking Flower will say:", elem_classes="wonder-card", elem_id="input_text")
    speak_button = gr.Button("Speak!", elem_id="comfirm_button", elem_classes="button wonder-card")
    audio_output = gr.Audio(label="输出音频", show_label=False, autoplay=True, elem_id="audio_output", elem_classes="wonder-card")
    
    speak_button.click(
        speak,
        inputs=[input_text],
        outputs=[audio_output]
    )


if __name__ == "__main__":
    hps = utils.get_hparams_from_file(config.webui_config.config_path)
    version = hps.version if hasattr(hps, "version") else latest_version
    net_g = get_net_g(
        model_path=config.webui_config.model, version=version, device=device, hps=hps
    )
    reload_javascript()
    demo.queue().launch(
        allowed_paths=["./assets"],
        show_api=False,
        # server_name=server_name,
        # server_port=server_port,
        share=True,
        inbrowser=True,  # 禁止在docker下开启inbrowser
    )