Sagnik Majumder

Papers

Total Citations

H-Index

About

Sagnik Majumder is a researcher at the forefront of embodied AI and audio-visual machine learning, with a focus on how intelligent agents perceive and interact with complex, real-world environments. His work bridges the gap between computer vision, audio processing, and robotics, tackling fundamental challenges in scene understanding and active perception. Majumder’s major contributions center on two key innovations. First, in his highly cited work *“Few-Shot Audio-Visual Learning of Environment Acoustics”* (17 citations), he pioneered methods for estimating room impulse responses (RIRs) from limited data, enabling AR/VR and robotic systems to realistically simulate how sound transforms in a physical space. Second, with *“Move2Hear: Active Audio-Visual Source Separation”* (4 citations), he introduced the novel problem of active source separation, where an agent must strategically move to isolate a target sound from a noisy, multi-source environment—a critical step toward autonomous navigation and hearing aids. His research demonstrates significant impact by redefining how machines leverage both sight and sound for spatial reasoning. By showing that an agent can learn to “move to hear,” Majumder has opened new avenues for interactive robotics, immersive virtual experiences, and assistive technologies. His work is a compelling read for any student interested in the intersection of perception, action, and multimodal learning.