Pocket Casts Mod
Ad-Free Podcast Pipeline
I made this tool because I was tired of ads cutting into every episode I listened to. It builds on MinusPod, an open-source project, and I customized it for Pocket Casts Plus by using its cloud files feature to sync the cleaned audio. Transcription runs locally with whisper.cpp, an open-source speech recognition engine, while ad detection uses a local Ollama model (a tool for running AI models locally) or a cloud AI service such as OpenRouter, OpenAI, or Groq. The Flask dashboard gives me a podcast-centric view with stat cards, per-episode queue controls, a backend services panel, and a searchable history of every episode I process, including time saved and ad counts.
My Contributions
I built this tool because I was tired of ads cutting into every episode I listened to. I started from MinusPod, an open-source project, and customized it for Pocket Casts Plus by using its cloud files feature to upload and sync the cleaned audio. I used AI assistance for the coding while I drove the architecture and design, integrated the Pocket Casts API for auth and episode listing, added a Flask dashboard with live log streaming and service controls, made the transcription backend portable across macOS and Linux with whisper.cpp, an open-source speech recognition engine, and used MinusPod patching with adjustable windows and padding. A searchable history tracks every episode I process, with ad counts and time saved.
Tools Used
Python, Flask, the Pocket Casts API, MinusPod, whisper.cpp, Ollama, and FFmpeg.