wtf is ddma?

DeepDive Media Automator, or DDMA, is a tool for turning long podcast episodes into short clips ready for TikTok, Instagram Reels, and YouTube Shorts. It combines a Python command line toolkit with an interactive web dashboard where you manage the whole process by hand. The workflow starts by dropping a podcast audio file into the project and creating a new project for it. The tool then transcribes the audio using OpenAI's Whisper model, producing a word-by-word transcript. From there, you use the dashboard to pick out the exact clips you want, clicking on individual words in the transcript to set precise start and end points down to a fraction of a second, and the interface highlights ranges you have already used so you do not accidentally overlap two clips. You can mix in background music for each clip segment, adjusting volume, crossfades, and duration, and manage a shared library of music files across projects. The dashboard also lets you switch between a few visual themes and remembers your choice. Once a clip is ready, DDMA can export it as an audio file or as a video with the audio laid over a plain colored background sized for square, vertical, or landscape formats, and it can automatically generate a title card at the start of the clip. There is also an optional integration with a service called Mosaic that can turn a finished draft into a polished video with AI generated infographic and motion graphic overlays, and the tool keeps track of these render jobs so it can pick up where it left off if the server restarts. Another optional feature uses Google's Gemini model to regenerate or recast an individual clip's script, using the style and pacing of previously finished clips as a guide. Running it requires Python, the FFmpeg video tool installed on your system, and the Whisper transcription library. The README recommends a dedicated NVIDIA graphics card with at least 4 GB of memory and 16 GB of system RAM, since transcription and video processing are demanding, though it can run on CPU alone at a much slower speed. No license information is given in the README.