[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f1wqa085cmu59v":3},{"code":4,"message":5,"data":6,"timestamp":57},10000,"success",{"full_name":7,"position":8,"description":9,"language":10,"stars":11,"forks":12,"stars_gained":13,"age_days":14,"hn_threads":15,"one_liner":28,"summary_md":29,"topic":30,"github_topics":33,"license":53,"homepage":54,"updated_at":55,"history":56},"abus-aikorea\u002Fvoice-pro",15,"Gradio WebUI for creators and developers, featuring key TTS (Edge-TTS, kokoro) and zero-shot Voice Cloning (E2 & F5-TTS, CosyVoice), with Whisper audio processing, YouTube download, Demucs vocal isolation, and multilingual translation.","Python",13349,1926,338,817,[16,22],{"id":17,"title":18,"points":19,"comments":20,"posted_at":21},42261909,"Show HN: Voice-Pro – AI Voice Cloning",271,189,"2024-11-28",{"id":23,"title":24,"points":25,"comments":26,"posted_at":27},42836934,"Show HN: Voice Cloning and Multilingual TTS in One Click (Windows)",9,3,"2025-01-27","本地 WebUI 把视频转写、翻译和克隆音色配音串成一条离线流水线。","## 解决什么问题\n把一段视频配成另一种语言，通常要拼下载器、人声分离、ASR、翻译、TTS 五六个独立工具，每个都有自己的安装和音频格式坑。创作者、播客和本地化研究者要么手工串脚本，要么按分钟付费给云端服务。Voice-Pro 把整条链路收进一个 Gradio 网页应用，从 YouTube 链接到成品配音都在本机完成。\n\n## 有什么不同\n常见做法是直接用 ElevenLabs 这类托管服务，或者自己用 Whisper 加某个 TTS 库搭。Voice-Pro 的差异在于一体化的覆盖面：识别端接 Whisper\u002FFaster-Whisper\u002FWhisper-Timestamped，翻译接 Deep-Translator，人声分离用 Demucs\u002FMDX-Net，TTS 同时提供 Edge-TTS、kokoro 和 F5-TTS、E2-TTS、CosyVoice 等零样本克隆模型。v4 把安装器从 Miniconda\u002Fpip 换成 uv 并锁定依赖，CUDA Toolkit 和 Visual Studio 构建工具也不再需要，因为依赖都有预编译 wheel。真正让它超出 demo 的是这层打包工作，而不是某一个模型。\n\n## 适合谁\n适合 Windows + NVIDIA GPU 的个人创作者、配音爱好者和小型本地化团队，想要离线、不订阅的工作流，并能接受模型下载体积和各语言音质参差。不适合需要 SLA 的高产量生产流水线，也不适合以 Linux\u002FMac 为主的团队：README 明说只验证过 Windows + NVIDIA，而且许可证描述自相矛盾（元数据写 GPL-3.0，文件头写 LGPL）。只缺某一个环节的开发者，直接用对应库更划算，没必要引入整个单体应用。\n\n## 上手成本\n需要 Windows、显存足够的 NVIDIA 显卡，以及装下一堆模型的磁盘空间，首次运行会从 Hugging Face 下载权重。翻译默认走免费的 Google 接口，可能被限流；可选的 Azure TTS 和翻译需要自备密钥。安装通过自带的 `start.bat` 完成，它会在本地拉起 uv 和 Python，具体见 README。最关键的是，README 说明作者因另一款产品暂停了维护，应当把它当成冻结的源码开放工具，而不是活跃维护的平台；再分发前请确认许可证文件。\n\n## 点评\nVoice-Pro 本质是一个打包得不错的集成层，而打包确实产生了效果：13.3k stars 和 2024 年底那个 271 分的 Show HN 帖说明有真实关注，2026 年 Python 周榜 +339 也说明 v4 重写重新带起了热度。但作者已声明暂停更新，风险是逐渐脱节：上游模型接口和 Gradio 版本会继续往前走，这个仓库未必跟得上。现在就需要离线配音 GUI、且是 Windows + N 卡的用户，它可以作为一站式选择；需要长期维护或脚本化流水线的人，更适合借用它的模型选型自己搭。",{"slug":31,"name":32},"media","多媒体与设计",[34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52],"audiobook","faster-whisper","gradio","karaoke","podcasts","speech-recognition","speech-synthesis","speech-to-text","subtitles","text-to-speech","transcription","translator","tts","voice-cloning","voice-conversion","webui","whisper","whisperx","yt-dlp","GPL-3.0","https:\u002F\u002Fwww.wctokyoseoul.com","2026-10-24T22:36:33Z",[],1792898632]