Relevant searches in podcasts with whisperx, pyannote, and RAG (FAISS and OpenAI API)

[pulijon]

The words are no longer lost to the wind. I’m proud to present Sttcast—a project that uses WhisperX for transcription, PyAnnote for speaker diarization, and FAISS and the OpenAI API for RAG (embeddings, summaries, context-driven Q&A). . Tested on a collection of 10 years of Podcasts (1,500 hours, more than 160 speakers). Truly amazed by the results. Sttcast