Generate images preserving face identity
Generate text from images and prompts
Generate speech from text using a reference voice