Speech and Multimodal Interfaces Laboratory

We received RSF grants as well as fellowship for PhD students and adjuncts

We received grants from the Russian Science Foundation (RSF):

  • RSF No 24-71-00083 «Research and Development of an Intelligent Gesture Recognition System for Controlling Human-Machine Interaction Interfaces»

    Head: Ryumin Dmitry
    Period: 2024-2026
    This scientific project aims to develop an intelligent system for recognizing human gestures, based on modern methods of deep machine learning. Gesticulation, an important element of non-verbal communication, plays a crucial role in communication and requires effective recognition to enhance the quality of human-machine interaction.

     

  • RSF No 24-71-00112 «Research and Development of a System for Synthesising Realistic Lip Movements of Digital Avatars According to Spoken Speech»

    Head: Axyonov Alexander
    Period: 2024-2026
    This scientific project aims to develop an intelligent system for synthesizing photorealistic lip movements of digital avatars, synchronized with input speech, using advanced machine learning methods. Realistic visualization of articulation plays a key role in creating expressive animated characters to improve the quality of human-computer interaction in many areas, including remote communications, education, entertainment, and others. The project represents an integration of modern deep neural network architectures, such as diffusion models, transformers, and convolutional networks, to build a highly accurate and adaptive lip motion synthesis system. The main tasks involve the development of mathematical, software, and informational support, divided into two stages: creating innovative models and software solutions, as well as conducting experimental research and thorough optimization.

     

In addition, Ryumina Elena was the recipient of a fellowship for PhD students and adjuncts with a research topic of «Automatic Evaluation of Human's Personality Traits by Multimodal Data. »
This scientific project is aimed at developing models and methods for the intelligent analysis of the five-factor model of personality using multimodal data (audio, video, and text). The five-factor model, also known as OCEAN, describes five personality traits: Openness to Experience, Conscientiousness, Extraversion, Agreeableness, and Non-Neuroticism. The project results will be applicable in various fields such as human resource management, marketing, education, and medicine, contributing to the improvement of the quality of life and professional performance.