Adobe has unveiled new generative AI audio tools designed to streamline video production for creators, enabling them to quickly add thematically matched soundtracks and voiceovers to their projects. The tools—Generate Soundtrack and Generate Speech—are now available in public beta within the redesigned Adobe Firefly AI app, while a new web-based video editor is in development to unify multiple AI-powered features into a single, intuitive interface. Generate Soundtrack analyzes an uploaded video and automatically generates a selection of instrumental audio clips that sync seamlessly with the footage. Users can choose from a range of style presets such as lofi, hip-hop, classical, and EDM, or describe the desired mood using a text prompt—requesting something more sentimental, energetic, or dramatic. The system also suggests a prompt based on the video’s content, helping users get started quickly. “We want to help users prompt music. It’s a new skill we need to develop, so we make it easier by predicting what type of music fits your clip,” said Alexandru Costin, Adobe’s head of generative AI. “We’re also offering a Mad Libs-style approach where you can pick the vibe, the style, and the objective.” Each prompt generates four distinct audio variations, with each clip capped at five minutes. The Firefly model behind the tool was trained exclusively on licensed music and voice content, ensuring that the generated audio is commercially safe and reduces the risk of copyright strikes—a key advantage over competitors like Suno and Udio, which have faced legal challenges over their use of protected material. “We purchased music and voice from IP owners, which is why we can confidently offer this as a safe, commercially viable solution,” Costin said. While Generate Soundtrack is designed for background music only, Adobe is actively developing additional tools to support a full AI-driven music production workflow, aiming to eliminate the legal and licensing hurdles that often frustrate creators. Also launching in public beta is Generate Speech, which creates voiceovers from text input. It offers over 50 voices powered by either Adobe’s Firefly Speech Model or ElevenLabs, supporting more than 20 languages. Users can adjust parameters like speech speed, pitch, and emotional tone, and manually correct pronunciation for names or words with regional variations. In parallel, Adobe is building a new web-based video editor called the Firefly video editor, described as a multitrack timeline tool for generating, organizing, trimming, and sequencing clips. It integrates AI features for voiceovers, soundtracks, and title generation, along with frame-by-frame editing and style presets. The tool will begin rolling out in private beta next month, with early access available only to those who sign up for a waitlist.