tongflow

Why we included this project

TongFlow gives you a visual canvas for linking text, image, video, music, and speech models into a single deliverable. A flow is built from three primitive operations: add an input, run it through a model, then combine the outputs, and each provider slots in as a drop-in node. The demo workflows show what that buys you: a talking-head video made from a script, a TTS voice, and a character image, or a music video assembled from lyrics, a generated song, and storyboard frames. Because it runs in the browser, can be self-hosted with Docker, and exports flows as portable JSON files, it suits small teams that want reproducible multimodal automation without a heavy framework. The plugin-based node library also keeps you on the providers you already pay for rather than locked into a single platform.