AI Pop-Up Video for OBS
A desktop companion app that listens to your stream / recording in real-time and triggers contextual pop-up overlays in OBS based on what you say.
While you talk, Google Cloud Speech-to-Text transcribes you live, GPT-4 watches for cue phrases or topics, and the app fires scene / source changes inside OBS via WebSocket — perfect for explainer-style streams, tutorials, sales demos, or character "companions" that react to your lines.
What's in the box
companion.py— main desktop app (Tkinter GUI). 655 lines. Connects to OBS WebSocket, records mic audio, streams it to GCP Speech-to-Text, calls OpenAI for reasoning + reply, drives OBS scenes/sources, renders the pop-up overlay.obs_handler.py— OBS WebSocket auth + scene switch helpers.audio_handler.py— PyAudio recording / chunking helpers.main.py— minimal "OBS Process Toggle" mini-app for quick scene switching.utils.py— logging + settings UI.*.png,*.svg— bundled overlay artwork (camcorder, comic bubbles, info bubbles).
