Videnza
AI ApplicationTurn any video into actionable intelligence. Paste a YouTube link or pick a local file — Videnza transcribes it, distills key insights, and lets you ask questions through a RAG-powered chatbot.
Videnza is a RAG-powered AI video assistant that transforms video content into structured intelligence. It supports YouTube URLs and local video/audio files, generating accurate transcriptions via Groq Whisper (English) and Sarvam AI (Hinglish). A map-reduce summarization pipeline powered by Mistral AI distills key insights, action items, and decisions, while a Pinecone vector store enables natural-language Q&A over the content. Results are surfaced through a clean dark-themed Streamlit UI with real-time pipeline progress tracking and PDF export for transcripts and insights.
Technologies Used
Key Features
- YouTube URL & local file upload support
- English transcription via Groq Whisper
- Hinglish transcription via Sarvam AI
- Map-reduce summarization with Mistral AI
- Automatic extraction of action items, key decisions, and open questions
- RAG-powered chatbot for querying video content
- Dark-themed Streamlit UI with pipeline progress tracking
- PDF export for transcripts and insight reports
- Audio processing pipeline with yt-dlp, FFmpeg, and pydub
- Vector-based retrieval using Pinecone and Mistral embeddings