Microsoft filed 3 patents this month across 2 categories: AI & Machine Learning (2) and Audio (1).
The AI & Machine Learning patents cover real-time voice transformation technology designed to match player voices to character appearances in role-playing games while preserving privacy and speech characteristics. The Audio patent describes real-time acoustic modeling for dynamic portals in games and VR environments, enabling precomputed sound propagation that adapts when virtual spaces change during gameplay.
Microsoft's 2 AI & Machine Learning patents both tackle the challenge of transforming player voices to match in-game characters during real-time multiplayer gameplay. The first system analyzes both player and character physical attributes through computer vision, then uses similarity matching algorithms to automatically select and blend voice characteristics that align with the character's appearance, eliminating the need for pre-recorded dialogue or manual voice acting. The second extends this concept by continuously adjusting vocal output based on character state, making voices sound muffled underwater or breathless while running, with synthesis updates occurring every frame to reflect changing game conditions while maintaining the speaker's natural speech patterns.
The single Audio patent addresses a longstanding limitation in game and VR sound design by enabling precomputed wave-based acoustic modeling to work with doors and openings that change state during gameplay. Traditional precomputed acoustic techniques only functioned in completely static environments, but this system stores portal path data and decouples portal attenuation from ambient acoustic parameters, allowing state changes to be applied efficiently at runtime without recalculating the entire acoustic simulation.
All data sourced from USPTO patent filings. Google Patents may take several weeks to index recent publications. If a link is unavailable, search for the patent number at USPTO Patent Public Search.