In the creation of captivating videos, audio plays a crucial role alongside visual effects. However, finding or producing suitable sound effects often proves to be a time-consuming task.
At Adobe's annual MAX conference, they showcased an experimental prototype named Project Super Sonic. This technology can generate sound effects from text, identify objects in videos, and even use your voice to quickly create background sound effects and audio for video projects.
While generating sound effects from text prompts sounds impressive, companies like ElevenLabs already offer similar services commercially. More intriguing is Adobe's advancement of this feature by introducing two additional ways to create soundtracks. The first involves an object recognition model where users can click on any part of a video frame, and the system will generate corresponding sound effect cues. This method of integrating multiple models into one workflow is quite intelligent.
The most astonishing feature is the third mode, where users can record their own mimicked sounds (synchronized with the video timeline), and Project Super Sonic will automatically generate suitable sound effects. Justin Salamon, the head of Adobe's sound design AI, mentioned that the team started with a text-to-audio model and emphasized that they only use licensed data in all Adobe generative AI projects.
"What we truly aim for is to empower users to control the entire process. This tool is designed for creators, sound designers, and anyone looking to enhance the audio quality of their videos," explained Salamon. "Thus, we are not just content with the initial text-to-sound effect workflow but are also developing a tool that offers precise control."
In terms of sound control, the tool actually analyzes various characteristics of the user's voice and the sound spectrum to guide the generation process. Salamon noted that although human voices were used in the demonstration, users can also clap their hands or use instruments for recording.
It's important to note that Adobe MAX always showcases so-called "Sneaks," experimental features like Project Super Sonic that Adobe is developing. While many of these projects eventually integrate into Adobe's creative suite, there's no guarantee that all will be officially launched. I believe Project Super Sonic has the potential to enter practical production, as the same team has also contributed to the audio part of Adobe Firefly's generative AI model, which can extend the duration of short videos, including their audio tracks. However, for now, Project Super Sonic remains a demonstration.
Key Points:
🎤 Project Super Sonic is an AI-powered experimental prototype that helps users quickly generate video sound effects.
🎶 Users can generate sound effects through text, video object recognition, and mimicked sounds, enhancing the creative experience.
🎥 Among the Sneaks projects showcased at Adobe MAX, Project Super Sonic is likely to be included in future creative suites.