Scrimba创始人Per在Hacker News上展示了一款新工具HN.watch，它能将Hacker News文章动态转换为AI生成的解说视频1。这项技术由Scrimba（获得YC S20融资的公司）开发，利用大语言模型在用户点击链接时实时创建视频内容1。
该工具具有显著的成本和性能优势1。每个视频的生成成本约为0.04美元（不含图像生成部分），从点击到播放仅需几秒1。相比基于扩散模型的方案，这一方案在速度、成本和可编辑性上都更具竞争力1。
HN.watch采用了多项关键技术支撑1。项目基于开源编程语言Imba构建，由Scrimba首席技术官Sindre Aarsæther创建1。生成过程中整合了Gemini、GPTs、Inworld和ElevenLabs等多个AI模型1。用户可以通过Web UI、MCP、ChatGPT Plugin以及Chrome Extension等多种方式使用该服务1。
Per, founder of Scrimba, unveiled HN.watch on Hacker News, a tool that automatically converts articles from the platform into video explanations using AI-powered video generation 1. The project leverages large language models to dynamically create videos whenever users click on links, delivering the content within seconds at an estimated cost of approximately $0.04 per video excluding image generation 1.
The solution employs a technology stack built on Imba, an open-source programming language developed by Scrimba's CTO Sindre Aarsæther 1. The video generation process integrates multiple AI services including Gemini, GPTs, Inworld, and ElevenLabs 1. Compared to diffusion-model-based approaches, the LLM-driven method offers advantages in speed, cost efficiency, and editability 1.
HN.watch is accessible through multiple interfaces to accommodate different user preferences 1. These include a web UI, an MCP integration, a ChatGPT plugin, and a Chrome extension 1. Scrimba previously received funding from Y Combinator's S20 cohort 1.
评论
还没有评论，欢迎留下第一条。