A person speaks in a normal dry room on land, but their voice sounds muffled, garbled, and distorted as if heard underwater — violating the physics of sound propagation in air, which should produce clear, undistorted speech.
Seedance 2.0ByteDance · proprietary
Kling 3.0 OmniKuaishou · proprietary
Veo 3.1Google · proprietary
LTX-2.3Lightricks · open
Ovi 1.1character.ai · open
JavisDiT++JavisDiT · open
MagiHumanGAIR-NLP · open
Per-prompt rubric
Basic standards
Visual presence
- a person
- an indoor room
Event: a person speaks in a dry indoor room on land
Audio presence
- a person
Sound: a person speaking
Key standards
Visual physical commonsense (V-PC)
- A person speaks in a visibly dry normal indoor room on land, with no water visible anywhere in the frame.
- The room and the person remain dry and on land throughout the clip, with no submersion or wet environment at any point.
Audio physical commonsense (A-PC)
- The voice sounds muffled, garbled, and distorted with the low-pass-filtered hollow character of speech transmitted through water rather than air.
Cross-modal physical commonsense (AV-PC)
- The underwater-character voice is present in the audio while the person is visibly speaking in the dry land-based room.