ADVANCING AUDIO-VISUAL SYNCHRONISATION
Join the Sounding Video Generation (SVG) Challenge 2024, a groundbreaking competition at the intersection of video generation and audio-visual synchronisation. This innovative challenge invites participants to build state-of-the-art AI models that generate perfectly aligned and contextually accurate videos guided by audio.
Use your machine learning expertise to advance the frontier of audio-visual synchronisation, transforming datasets into dynamic, synchronised videos. With the right code and creativity, you’ll contribute to a rapidly evolving field that’s set to redefine how we generate and experience multi-modal content.
The Task
Develop models that generate videos with synchronised and contextually relevant audio in two specialised tracks:
- Temporal Alignment Track: Create videos where the audio is perfectly synchronised with the video content in time (e.g., a dog barking exactly when seen in the video).
- Spatial Alignment Track: Develop models that produce videos with spatially aligned audio, creating a real sense of direction and space.
Both tracks aim to push the boundaries of multi-modal AI, offering a unique platform to benchmark cutting-edge solutions in this underexplored domain.
Timeline
- Warm-Up Round: Starts 1st October 2024
- Phase I: Begins 21st October 2024
- Phase II: Starts 2nd December 2024
- Challenge End: 31st January 2025
Prizes
The challenge boasts a prize pool of USD 35,000 split across both tracks. The top three teams or participants in each track will be rewarded as follows:
-
Track 1: Temporal Alignment ($17,500)
-
First place: USD 10,000 -
Second place: USD 5,000 -
Third place: USD 2,500
-
-
Track 2: Spatial Alignment ($17,500)
-
First place: USD 10,000
-
Second place: USD 5,000
-
Third place: USD 2,500
-
Join a Thriving Community
Collaborate with like-minded researchers, practitioners, and AI enthusiasts eager to share ideas, team up, and drive innovation in this captivating challenge.
Sign up now for the SVG Challenge 2024 and start building your models in the warm-up round!