What is e3d-pod2vid?

e3d-pod2vid is an AI powered pipeline written in Python and Node.js that turns an existing audio recording, such as a podcast, an interview, or an export from NotebookLM, into a finished video ready to upload to YouTube. The core script sends the audio to AssemblyAI to figure out who is speaking and when, asks a small AI model to pick a relevant stock video clip from Pexels for each spoken segment, downloads and caches those clips, then renders the final video with subtitles burned directly into the picture using the Pillow image library rather than needing separate subtitle software installed. Beyond that main conversion, the toolkit includes several optional extra steps. One script can replace the original voices in the audio with AI generated text to speech voices from OpenAI, useful for swapping out a synthetic NotebookLM voice for a more natural sounding one. Another script builds short, vertical format videos formatted for YouTube Shorts, Instagram Reels, or TikTok, complete with a narrated script, matching stock footage, subtitles, and a closing call to action card with a logo. A separate automated script can check a live signals feed from a companion service called maps.e3d.ai, automatically pick the most notable signal of the day based on rules like confidence level or risk, generate a short video about it, and publish it, intended to run on a daily schedule. Once a video is rendered, separate scripts handle uploading it to YouTube with a title, description, and thumbnail, and then announcing the new video across several social platforms at once, including Discord, Telegram, X, LinkedIn, and a platform called Moltbook, skipping any platform for which credentials were not provided. Running the pipeline requires API keys for services like AssemblyAI, OpenAI, and the individual social platforms you want to use, since each step depends on an outside AI or publishing service rather than running entirely offline. There is also a separate wrapper script meant for running the pipeline as a hosted background job, reading a structured job description and writing its progress out as it works.