Open to Collaborate -- HORIZON-CL4-2025-03-HUMAN-14: Core technologies for virtual worlds

Innovation Action SESTEK can Contribute:

A. Better and more realistic immersion through multimodal interaction. Improve the multimodal immersion experience combining eXtended Reality with advanced and innovative technologies.

• AI-powered assistants can understand natural language commands, making XR environments more user-friendly.

• Users can speak naturally to control virtual elements, reducing reliance on controllers or touch-based inputs.

SESTEK’s Text-to-Speech (TTS) technology can make virtual assistants and NPCs (non-player characters) sound human-like and expressive.

• SESTEK’s emotion detection can detect voice tone, stress, and sentiment, enabling virtual assistants to adapt responses based on user emotions.

• The low-code business flow design enables easy customization for various industry verticals without third-party dependencies.

• Knovvu Virtual Agent enhances personalization through RAG Semantic Caching, which optimizes response time and cost efficiency by storing and reusing previously generated answers for similar queries.

• Personalization via Memory

o Short-Term Memory: Retains context during a session, ensuring coherent and logical conversations.

o Long-Term Memory: Remembers user-specific details like preferences and facts from past interactions, offering a hyper-personalized experience in future engagements.

• Personalized AI can help virtual characters in XR environments respond more naturally to user emotions.

• Speech solutions can be integrated with haptic feedback devices, making interactions in XR more immersive.

• For example, a virtual assistant could guide users through repair procedures, providing step-by-step spoken instructions with synchronized haptic feedback.

Virtual Agent:

SESTEK’s AI-powered Knovvu Virtual Agent (TRL 8-9) excels in accurately recognizing user utterances and converting spoken words into written text. With a market-leading AI-based intent recognition accuracy of over 96%, it ensures precise and efficient interactions. As an all-in-one solution, it integrates Speech Recognition (SR), Text-to-Speech (TTS), Natural Language Understanding (NLU), an orchestrator, and a design studio eliminating the need for third-party involvement. We also benefit from both open source and API based Generative models into the architecture. Key Features: https://docs.knovvu.com/docs/features