| top_md_1 = (""" | |
| <div align="center"> | |
| <div style="display:flex; gap: 0.25rem;" align="center"> | |
| FunClip: <a href='https://github.com/alibaba-damo-academy/FunClip'><img src='https://img.shields.io/badge/Github-Code-blue'></a> | |
| 🌟支持我们: <a href='https://github.com/alibaba-damo-academy/FunClip/stargazers'><img src='https://img.shields.io/github/stars/alibaba-damo-academy/FunClip.svg?style=social'></a> | |
| </div> | |
| </div> | |
| 基于阿里巴巴通义实验室自研并开源的[FunASR](https://github.com/alibaba-damo-academy/FunASR)工具包及Paraformer系列模型及语音识别、端点检测、标点预测、时间戳预测、说话人区分、热词定制化开源链路 | |
| 准确识别，自由复制所需段落，或者设置说话人标识，一键裁剪、添加字幕 | |
| * Step1: 上传视频或音频文件（或使用下方的用例体验），点击 **<font color="#f7802b">识别</font>** 按钮 | |
| * Step2: 复制识别结果中所需的文字至右上方，或者右设置说话人标识，设置偏移与字幕配置（可选） | |
| * Step3: 点击 **<font color="#f7802b">裁剪</font>** 按钮或 **<font color="#f7802b">裁剪并添加字幕</font>** 按钮获得结果 | |
| 🔥 FunClip现在集成了大语言模型智能剪辑功能，选择LLM模型进行体验吧~ | |
| """) | |
| top_md_2 = ("通过Modelscope创空间的访问可能由于多人同时使用的原因不稳定，推荐通过github项目中的代码自己部署。") | |
| top_md_3 = ("""访问FunASR项目与论文能够帮助您深入了解ParaClipper中所使用的语音处理相关模型： | |
| <div align="center"> | |
| <div style="display:flex; gap: 0.25rem;" align="center"> | |
| FunASR: <a href='https://github.com/alibaba-damo-academy/FunASR'><img src='https://img.shields.io/badge/Github-Code-blue'></a> | |
| FunASR Paper: <a href="https://arxiv.org/abs/2305.11013"><img src="https://img.shields.io/badge/Arxiv-2305.11013-orange"></a> | |
| 🌟Star FunASR: <a href='https://github.com/alibaba-damo-academy/FunASR/stargazers'><img src='https://img.shields.io/github/stars/alibaba-damo-academy/FunASR.svg?style=social'></a> | |
| </div> | |
| </div> | |
| """) | |
| top_md_4 = ("""我们在「LLM智能裁剪」模块中提供三种LLM调用方式， | |
| 1. 选择阿里云百炼平台通过api调用qwen系列模型，此时需要您准备百炼平台的apikey，请访问[阿里云百炼](https://bailian.console.aliyun.com/#/home)； | |
| 2. 选择GPT开头的模型即为调用openai官方api，此时需要您自备sk与网络环境； | |
| 3. [gpt4free](https://github.com/xtekky/gpt4free?tab=readme-ov-file)项目也被集成进FunClip，可以通过它免费调用gpt模型； | |
| 其中方式1与方式2需要在界面中传入相应的apikey | |
| 方式3而可能非常不稳定，返回时间可能很长或者结果获取失败，可以多多尝试或者自己准备sk使用方式1 | |
| """) | |
| top_md_1_en = (""" | |
| Powered by Alibaba DAMO Academy's in-house developed and open-sourced [FunASR](https://github.com/alibaba-damo-academy/FunASR) toolkit and the Paraformer series models, offering open link to voice recognition, endpoint detection, punctuation prediction, timestamp prediction, speaker differentiation, and customization of hot words. | |
| Accurate recognition, freely copy the required paragraphs, or set speaker labels, one-click clipping, adding subtitles | |
| * Step1: Upload a video or audio file (or use the case below for a demo), then click **<font color="#f7802b">Recognize</font>** button | |
| * Step2: Copy the required text from the recognition results to the top right, or set speaking indicators on the right and configure offset and subtitle settings (optional) | |
| * Step3: Click the **<font color="#f7802b">Clip</font>** button or **<font color="#f7802b">Clip and Add Subtitles</font>** button to get the result | |
| 🔥 FunClip now integrates the smart editing functionality with large language models, choose an LLM model for a try~ | |
| """) | |
| top_md_2_en = (""" | |
| Access to the HuggingFace space may be unstable due to simultaneous use by multiple users. It is recommended to deploy the code through the GitHub project yourself. | |
| <div align="center"> | |
| <div style="display:flex; gap: 0.25rem;" align="center"> | |
| FunClip: <a href='https://github.com/alibaba-damo-academy/FunClip'><img src='https://img.shields.io/badge/Github-Code-blue'></a> | |
| 🌟 Support Us: <a href='https://github.com/alibaba-damo-academy/FunClip/stargazers'><img src='https://img.shields.io/github/stars/alibaba-damo-academy/FunClip.svg?style=social'></a> | |
| </div> | |
| </div> | |
| """) | |
| top_md_3_en = ("""Visiting the FunASR project and paper can help you gain an in-depth understanding of the speech processing models used in ParaClipper: | |
| <div align="center"> | |
| <div style="display:flex; gap: 0.25rem;" align="center"> | |
| FunASR: <a href='https://github.com/alibaba-damo-academy/FunASR'><img src='https://img.shields.io/badge/Github-Code-blue'></a> | |
| FunASR Paper: <a href="https://arxiv.org/abs/2305.11013"><img src="https://img.shields.io/badge/Arxiv-2305.11013-orange"></a> | |
| 🌟Star FunASR: <a href='https://github.com/alibaba-damo-academy/FunASR/stargazers'><img src='https://img.shields.io/github/stars/alibaba-damo-academy/FunASR.svg?style=social'></a> | |
| </div> | |
| </div> | |
| """) | |
| top_md_4_en = ("""We provide three ways to use LLMs in the 'LLM Clipping' module: | |
| 1. Choose Alibaba Cloud's BaiLian platform to call qwen series models via API, for which you need to prepare the BaiLian platform's apikey. Please visit [Alibaba Cloud BaiLian](https://bailian.console.aliyun.com/#/home); | |
| 2. Choosing models that start with GPT denotes the use of the official OpenAI API, for which you need to provide your own sk and network environment; | |
| 3. The [gpt4free](https://github.com/xtekky/gpt4free?tab=readme-ov-file) project is also integrated into FunClip, allowing free use of the gpt model; | |
| Both methods 1 and 2 require you to enter the respective apikey in the interface. | |
| Method 3 may be highly unstable, with potential long return times or failure to retrieve results. You may try multiple times or prepare your own sk to use method 1. | |
| """) | |