3 ms·
VideoWise – Make Video Content Searchable and Interactive with AI
- GioPicci 2y agoHave you ever spent hours scrubbing through a 3-hour Zoom recording, desperately searching for that one critical piece of information? It’s frustrating, time-consuming, and (mostly) avoidable. I’ve been working on VideoWise to solve exactly this problem. It's an open-source app designed to make video content easier to navigate, search, and use. It’s something I originally built to help my coworkers, who often had to sift through hours of recordings to learn or remember specific procedures. Since most of these videos contain sensitive corporate data, I made sure everything runs locally to maintain privacy. Here’s what VideoWise does right now: - Automatic Transcriptions: It uses WhisperX (based on OpenAI's Whisper) to transcribe videos, with timestamps for every sentence. This makes it easy to jump between sections of the video. - Automatic Translations: WhisperX also supports translating the video content. Sometimes it can mess up and skip parts of the subtitles but, in general, it works pretty well and can be really useful in the right situations. - AI-Powered Interaction: Integrates with Ollama, so you can ask questions, get summaries, or generate content like quizzes. Since Ollama is run locally, there’s no need to send private data to services like ChatGPT or Claude. - Export Options: Save transcriptions with subtitles or download AI-generated chats in various formats for later use. Of course, most of this stuff can be accomplished by using a combination of already existing tools, but I think there's some value in having a single app that does all of the above without relying on external service. It’s fully open-source and modular—you can deploy everything with Docker or break it into separate services for scalability. I’m hoping it’ll be useful for anyone working with large volumes of video, whether for training, education, or content creation. I’d love feedback from the HN community. Are there features you think would be valuable? How could I make it more useful? Thanks for checking it out!
- toomuchtodo 2y agoGreat work! Based on your TODO "Add functionality to upload videos directly from YouTube URLs." I recommend leveraging https://github.com/yt-dlp/yt-dlp https://github.com/yt-dlp/yt-dlp to support any video source their extractors support. Also consider S3 compatible targets (not just AWS S3, but also Backblaze B2, Minio, etc). Lastly, consider supporting writing post processing data back to the system(s). This should cover ingesting data from the vast majority of potential use cases besides a local filesystem mount.
- GioPicci 2y agoThank you so much for your suggestions! Leveraging yt-dlp is a fantastic idea—it’ll make supporting YouTube URLs and other video sources much easier, so I’ll definitely prioritize adding that. Supporting S3-compatible targets is also something I’d love to implement since it would give users a lot more flexibility with their workflows. I'll definitely look into that. Great feedback!