Overview

• Firefly’s Generate Sound Effects beta produces synchronized custom audio from text descriptions and onomatopoeic voice recordings within video timelines.
• New video controls include Composition Reference for mirroring reference footage, Keyframe Cropping for defining start and end frames, and Style Presets for one-tap visual themes.
• Text to Avatar beta turns scripts into avatar-led videos with customizable backgrounds and accent options.
• The platform now integrates external models such as Moonvalley’s Marey, Google’s Veo 3 with audio, Pika 2.2, Luma AI’s Ray 2 and Topaz Labs’ upscalers under a single Adobe sign-in.
• Adobe reiterates that Firefly is trained only on licensed or public-domain data and does not use users’ content to train its generative models.