Spaces:
Paused
Paused
| import os | |
| import tempfile | |
| from openai import OpenAI | |
| from tts_voice import tts_order_voice | |
| import edge_tts | |
| import numpy as np | |
| import anyio | |
| import torch | |
| import torchaudio | |
| import gradio as gr | |
| from scipy.io import wavfile | |
| from scipy.io.wavfile import write | |
| #新加内容 | |
| import asyncio | |
| import threading | |
| import requests | |
| from aiohttp import ClientSession | |
| # # 异步函数进行预加载 | |
| # async def fetch_link_content(url): | |
| # async with ClientSession() as session: | |
| # async with session.get(url) as response: | |
| # return await response.text() | |
| # # 后台任务确保不阻塞主线程 | |
| # def fetch_link_in_background(url): | |
| # loop = asyncio.new_event_loop() | |
| # asyncio.set_event_loop(loop) | |
| # content = loop.run_until_complete(fetch_link_content(url)) | |
| # # 将 content 缓存起来或者在全局状态中保存以供后续使用 | |
| # print("预加载的内容:", content) | |
| # link_url = "https://huggingface.co/api/spaces/by-subdomain/zxsipola123456-tts" | |
| # background_thread = threading.Thread(target=fetch_link_in_background, args=(link_url,)) | |
| # background_thread.start() | |
| # 创建 KNN-VC 模型 | |
| knn_vc = torch.hub.load('bshall/knn-vc', 'knn_vc', prematched=True, trust_repo=True, pretrained=True, device='cpu') | |
| # 初始化 language_dict | |
| language_dict = tts_order_voice | |
| # 异步文字转语音函数 | |
| async def text_to_speech_edge(text, language_code): | |
| voice = language_dict[language_code] | |
| communicate = edge_tts.Communicate(text, voice) | |
| with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as tmp_file: | |
| tmp_path = tmp_file.name | |
| await communicate.save(tmp_path) | |
| return "语音合成完成：{}".format(text), tmp_path | |
| def voice_change(audio_in, audio_ref): | |
| samplerate1, data1 = wavfile.read(audio_in) | |
| samplerate2, data2 = wavfile.read(audio_ref) | |
| with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_audio_in, \ | |
| tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_audio_ref: | |
| audio_in_path = tmp_audio_in.name | |
| audio_ref_path = tmp_audio_ref.name | |
| write(audio_in_path, samplerate1, data1) | |
| write(audio_ref_path, samplerate2, data2) | |
| query_seq = knn_vc.get_features(audio_in_path) | |
| matching_set = knn_vc.get_matching_set([audio_ref_path]) | |
| out_wav = knn_vc.match(query_seq, matching_set, topk=4) | |
| output_path = 'output.wav' | |
| torchaudio.save(output_path, out_wav[None], 16000) | |
| return output_path | |
| # #验证中转api key是否有效 | |
| # def validate_api_key(api_proxy_key): | |
| # try: | |
| # client = OpenAI(api_key=api_proxy_key, base_url='https://lmzh.top/v1') | |
| # # 测试调用一个简单的API来验证Key | |
| # response = client.models.list() | |
| # return True | |
| # except Exception: | |
| # return False | |
| # # 更新Edge TTS标签页状态的函数 | |
| # def update_edge_tts_tab(api_proxy_key): | |
| # is_valid = validate_api_key(api_proxy_key) | |
| # return gr.update(interactive=is_valid) | |
| # 文字转语音（OpenAI） | |
| def tts(text, model, voice, api_key): | |
| if len(text) > 300: | |
| raise gr.Error('您输入的文本字符多于300个，请缩短您的文本') | |
| if api_key == '': | |
| raise gr.Error('请填写您的 中转API Key') | |
| try: | |
| client = OpenAI(api_key=api_key, base_url='https://lmzh.top/v1') | |
| response = client.audio.speech.create( | |
| model=model, | |
| voice=voice, | |
| input=text, | |
| ) | |
| except Exception as error: | |
| raise gr.Error(f"生成语音时出错：{error}") | |
| with tempfile.NamedTemporaryFile(suffix=".mp3", delete=False) as temp_file: | |
| temp_file.write(response.content) | |
| temp_file_path = temp_file.name | |
| return temp_file_path | |
| def tts1(text, model, voice, api_key): | |
| if len(text)>300: | |
| raise gr.Error('您输入的文本字符多于300个，请缩短您的文本') | |
| if api_key == '': | |
| raise gr.Error('Please enter your OpenAI API Key') | |
| else: | |
| try: | |
| client = OpenAI(api_key=api_key) | |
| response = client.audio.speech.create( | |
| model=model, # "tts-1","tts-1-hd" | |
| voice=voice, # 'alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer' | |
| input=text, | |
| ) | |
| except Exception as error: | |
| # Handle any exception that occurs | |
| raise gr.Error("An error occurred while generating speech. Please check your API key and try again.") | |
| print(str(error)) | |
| # Create a temp file to save the audio | |
| with tempfile.NamedTemporaryFile(suffix=".mp3", delete=False) as temp_file: | |
| temp_file.write(response.content) | |
| # Get the file path of the temp file | |
| temp_file_path = temp_file.name | |
| return temp_file_path | |
| # Gradio 前端设计 | |
| app = gr.Blocks(title="TTS文本生成语音 + AI秒变声") | |
| with app: | |
| gr.Markdown("# <center>TTS文本生成语音 + AI秒变声</center>") | |
| gr.Markdown("### <center>key获取地址[here](https://buy.sipola.cn),ai文案生成(可使用中转key和官方key)请访问 [here](https://ai.sipola.cn)</center>") | |
| with gr.Tab("中转key-TTS文本生语音"): | |
| with gr.Row(variant='panel'): | |
| api_proxy_key = gr.Textbox(type='password', label='API Key', placeholder='请在此填写您在https://buy.sipola.cn获取的中转API Key') | |
| model = gr.Dropdown(choices=['tts-1','tts-1-hd'], label='请选择模型（tts-1推理更快，tts-1-hd音质更好）', value='tts-1') | |
| voice = gr.Dropdown(choices=['alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer'], label='请选择一个说话人', value='alloy') | |
| with gr.Row(): | |
| with gr.Column(): | |
| inp_text = gr.Textbox(label="请填写您想生成的文本中英文皆可", placeholder="请输入ai生成的文案，不要超过300字，最好200字左右", lines=5) | |
| btn_text = gr.Button("一键生成音频", variant="primary") | |
| with gr.Column(): | |
| inp1 = gr.Audio(type="filepath", label="TTS真实拟声", interactive=False) | |
| inp2 = gr.Audio(type="filepath", label="请上传同文案参照音频，可自己读取同文案录音") | |
| btn1 = gr.Button("一键AI变声合成", variant="primary") | |
| with gr.Column(): | |
| out1 = gr.Audio(type="filepath", label="AI变声后的专属音频") | |
| btn_text.click(tts, [inp_text, model, voice, api_proxy_key], inp1) | |
| btn1.click(voice_change, [inp1, inp2], out1) | |
| with gr.Tab("官方key-TTS文本生语音"): | |
| with gr.Row(variant='panel'): | |
| api_key = gr.Textbox(type='password', label='API Key', placeholder='请在此填写您在https://buy.sipola.cn 获取的官方API Key') | |
| model = gr.Dropdown(choices=['tts-1','tts-1-hd'], label='请选择模型（tts-1推理更快，tts-1-hd音质更好）', value='tts-1') | |
| voice = gr.Dropdown(choices=['alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer'], label='请选择一个说话人', value='alloy') | |
| with gr.Row(): | |
| with gr.Column(): | |
| inp_text = gr.Textbox(label="请填写您想生成的文本中英文皆可", placeholder="请输入ai生成的文案，不要超过300字，最好200字左右", lines=5) | |
| btn_text = gr.Button("一键生成音频", variant="primary") | |
| with gr.Column(): | |
| inp1 = gr.Audio(type="filepath", label="TTS真实拟声", interactive=False) | |
| inp2 = gr.Audio(type="filepath", label="请上传同文案参照音频，可自己读取同文案录音") | |
| btn1 = gr.Button("一键AI变声合成", variant="primary") | |
| with gr.Column(): | |
| out1 = gr.Audio(type="filepath", label="AI变声后的专属音频") | |
| btn_text.click(tts1, [inp_text, model, voice, api_key], inp1) | |
| btn1.click(voice_change, [inp1, inp2], out1) | |
| with gr.Tab("TTS-AI变声") as edge_tts_tab: | |
| with gr.Row(): | |
| with gr.Column(): | |
| input_text = gr.Textbox(label="请填写您想生成的文本中英文皆可",placeholder="请输入ai生成的文案，不要超过300字，最好200字左右",lines=5) | |
| btn_edge = gr.Button("一键生成音频", variant="primary") | |
| with gr.Column(): | |
| default_language = list(language_dict.keys())[15] | |
| language = gr.Dropdown(choices=list(language_dict.keys()), value=default_language, label="请选择文本对应的语言") | |
| output_audio = gr.Audio(type="filepath", label="TTS真实拟声", interactive=False, show_download_button=False) | |
| output_text = gr.Textbox(label="输出文本", visible=False) | |
| with gr.Row(): | |
| with gr.Column(): | |
| inp_vc = gr.Audio(type="filepath", label="请上传和文案相同参照音频，可自己读取文案录音或者用TTS文本生语音同文案生成的音频") | |
| btn_vc = gr.Button("一键AI变声合成", variant="primary") | |
| with gr.Column(): | |
| out_vc = gr.Audio(type="filepath", label="AI变声后的专属音频") | |
| btn_edge.click(lambda text, lang: anyio.run(text_to_speech_edge, text, lang), [input_text, language], [output_text, output_audio]) | |
| btn_vc.click(voice_change, [output_audio, inp_vc], out_vc) | |
| # 监听API Key输入框的变化并更新Edge TTS标签页的状态 | |
| # api_proxy_key.change( | |
| # update_edge_tts_tab, | |
| # inputs=[api_proxy_key], | |
| # outputs=[edge_tts_tab] | |
| # ) | |
| gr.HTML(''' | |
| <div class="footer"> | |
| <center><p>Power by sipola </p></center> | |
| </div> | |
| ''') | |
| app.launch(show_error=True) | |