Self-hosted podcast search stack: Whisper + pyannote + pgvector indexes 180 hours
Apprehensive-Ear9499 · reddit · 2026-09-09
A developer open-sourced podcast-mcp, a self-hostable stack that turns any podcast into a searchable, chat-queryable database — 180 hours indexed with a live read-only demo.
- Pipeline: ingest podcast RSS → Whisper transcription → pyannote speaker diarization → LLM-based speaker-name resolution → chunking + OpenAI embeddings → Postgres/pgvector with timestamps.
- Exposed as an MCP server, so Claude, Cursor, or any MCP client can ask things like "what was said about X across all episodes?" and jump back to exact timestamps.
- Runs via Docker Compose; GPU transcription can offload to RunPod, so a small VPS suffices. Works with any RSS feed. Repo: github.com/bramdehart/podcast-mcp
More from coding & agent
- Fresh Memory, Stale Plans: PLANFENCE Blocks All Invalid Agent Actions Across 30 Workflows — rohanpaul_ai · 2026-09-09
- VoiceStudio open-sources a local ElevenLabs-style stack: voice clone, dubbing, 646 languages — thisguyknowsai · 2026-09-09
- Teknium says missing /review subagent is back — update to restore it — Teknium · 2026-09-09
- Layered summaries keep context at 8k, yet Qwen degrades after ~120 chat turns — Zeeplankton · 2026-09-09
- IndeHub Chapter 16: hands-on session on building, shipping and monetizing apps with AI agents — rudrank · 2026-09-09
- Users want AI assistants to talk to each other, but lock-in makes it unlikely — mobileraj · 2026-09-09