Removes ads from podcasts using Whisper + AI. Serves clean RSS feeds for any podcast app.
100K+
MinusPod is a self-hosted server that strips ads from podcasts and YouTube shows before you hit play. Point it at an RSS feed, playlist or channel and subscribe to the feed it serves. Podcasts are transcribed locally (faster-whisper, Parakeet or a remote API) and an LLM cuts the ads. Your corrections become patterns that catch repeat sponsors without another LLM call. YouTube audio comes through yt-dlp and SponsorBlock marks decide the cuts. Bring your own LLM: Claude, Ollama, OpenRouter, OpenAI-compatible providers or System One.
Website: minuspod.app. Documentation: docs.minuspod.app. Demo: demo.minuspod.app.
A tour of the web UI. More videos, portrait cuts and the files themselves in docs/videos.
https://github.com/user-attachments/assets/2078724e-ca79-4368-a901-3817f9a0d213
Ad detection
Transcription and LLMs
Patterns and sponsors
Publishing
YouTube
Web UI and administration
The transcriber turns the episode into text, an LLM marks the ad segments, FFmpeg cuts them, and Flask serves the modified feed and audio. One episode, end to end:
https://github.com/user-attachments/assets/cf3d38d7-3df6-4e8d-bac1-c5dd720dc4b6
The full pipeline, including the verification pass, sliding windows, pattern learning, and audio analysis, is in docs/how-it-works.md.
Whisper and Parakeet sizing and RAM/VRAM figures are in Installation.
# 1. Create environment file
cat > .env << EOF
ANTHROPIC_API_KEY=your-key-here
BASE_URL=http://localhost:8000
MINUSPOD_MASTER_PASSPHRASE=long-random-string-you-will-not-lose
OPENAI_MODEL=claude-haiku-4-5
EOF
# 2. Create data directory
mkdir -p data
# 3. Run (pick one)
docker compose up -d # NVIDIA GPU image
docker compose -f docker-compose.cpu.yml up -d # CPU image, no GPU needed
Open http://localhost:8000/ui/, set an application password, and add a feed. The GPU image is amd64 with the NVIDIA container runtime; the CPU image runs natively on amd64 and arm64. Both read the same .env and data/ directory.
MINUSPOD_MASTER_PASSPHRASE encrypts provider keys at rest, so set it before you store any. Remote first-time setup, public processing, and the slower stable release channel are covered in Installation.
| Topic | |
|---|---|
| Installation | Requirements, sizing, quick start, CPU image |
| Environment variables | Every env var |
| LLM providers | Providers, recommended models, pricing |
| Transcription | faster-whisper, Parakeet, remote APIs, OpenVINO |
| Web interface | Management UI and the ad editor |
| How it works | The detection pipeline in depth |
| Workflows | Processing modes, re-runs, failure paths |
| Configuration | Every setting and experiment |
| Security and storage | Remote access, authenticated feeds, backups |
| API and webhooks | REST endpoints, webhook events, notifications |
| Podcasting 2.0 | What is emitted, regenerated, and stripped |
| Glossary | Every term the app uses |
The full docs index covers the rest: audio cues, community patterns, Pattern Cleanup, System One, failover, local feeds, YouTube feeds, download history, configuration import and export, deployment, releasing, and the LLM benchmark report.
Bug reports, feature requests, and pull requests are welcome. Open an issue first for larger changes so we can agree on the approach. See the contributor guide for development setup, build instructions, and tests. Ad patterns go through Community Patterns and benchmark results through Benchmark Contributing. Fork-derived features and community bug reports are credited in the CHANGELOG entry that ships them.
This tool is for personal use only. Only use it with podcasts you have permission to modify or where such modification is permitted under applicable laws. Respect content creators and their terms of service.
YouTube feeds download audio with yt-dlp. That can conflict with YouTube's terms of service; the feature is off by default and is yours to enable.
MIT
This project was developed using AI agents as a pair programmer. It was NOT vibe coded. For context, I'm a systems engineer who also writes code professionally with 15+ years of experience. The codebase follows engineering best practices, and all architecture and design decisions were made by me, not by AI. All code generated by LLMs was reviewed and tested by me, a human.
Content type
Image
Digest
sha256:b5cfa4ece…
Size
2.6 GB
Last updated
about 13 hours ago
docker pull ttlequals0/minuspod