DramaBox by Resemble AI

About

A text-to-speech model ought to provide two essential elements: an award-caliber performance and a certified signature to confirm its authenticity. DramaBox is the pioneering model to deliver both. You can direct a scene as you would instruct an actor, for instance by saying "a talk show host gasps in mock shock, then bursts into laughter," and the model will translate that into its performance. Each audio result is secured with the Resemble Watermarker. The model is open source and currently supports only English. You can access it through your Resemble account or on Hugging Face.