Deep-learning based prediction of Energy Decay Curves and Room Impulse Response reconstruction

#roomimpulseresponse #acoustics #signalprocessing Deep-learning based prediction of Energy Decay Curves (EDCs) and Room Impulse Response (RIR) reconstruction - Dr. Imran Muhammad 🔊 New Series Announcement | Audio Technology, Acoustics & Music Research The BRICS+ Audio, Acoustics & Music Tech Group, in cooperation with The Sound Travels YouTube channel, is launching a new video series dedicated to showcasing research in audio technology, acoustics, and music technology. This initiative aims to amplify the voices of scientists, engineers, and researchers; particularly from BRICS+ countries and the wider Global South, including Africa, Latin America, and Asia; whose work in sound, audio, and music-related technologies is shaping innovation but remains underrepresented in global platforms. 🎧 Focus areas include (but are not limited to): • Audio signal processing • Architectural and environmental acoustics • Music technology and digital instruments • Psychoacoustics and perception • Sound studies and emerging audio technologies 🎥 What the series offers: • Short, accessible presentations of ongoing or completed research • A platform to share ideas with a global, interdisciplinary audience • Visibility for research addressing local and global sonic challenges Researchers at all career stages are welcome. If you are interested in participating;or know someone whose work should be featured; we invite you to get in touch: leave a comment or DM us. Let’s connect sound, science, and innovation across the Global South and beyond. --- Abstract: This work proposes a neural network framework for efficient room impulse response (RIR) prediction by estimating energy decay curves (EDCs) from room geometry, material absorption, and source–receiver positions, followed by RIR reconstruction via reverse-differentiation. Trained on large-scale simulated datasets with realistic acoustic conditions, the method achieves high accuracy in objective metrics and shows no perceptual differences from reference RIRs in MUSHRA listening tests. The results demonstrate a fast, accurate, and perceptually reliable alternative to conventional room acoustic simulation methods for immersive audio applications. Link: https://github.com/TUIlmenauAMS/LSTM-Model-Energy-Decay-Curves Bio Dr. Imran Muhammad is an acoustics researcher and educator with a background in Electrical Engineering and Information Technology from RWTH Aachen University, Germany. He is currently an Associate at the Technical University Ilmenau, where he teaches master’s courses and supervises thesis projects. His expertise includes room acoustics, acoustics for virtual reality, audio / video coding and Machine Learning. He began his research career at Hanyang University in Seoul, Korea, contributing to several projects on virtual acoustic rendering and earning four patents in sound synthesis, and spatial audio. He completed his Ph.D. at the Institute of Technical Acoustics, RWTH Aachen, focusing on building acoustic auralization for psychoacoustic experiments, with multiple journal publications. He has also held research and supervision roles at the Technical University Eindhoven and worked as a Research Engineer at Brandenburg Labs, focusing on interactive auralization and immersive acoustic environments. More about ME: https://imran.virtual-acoustics.org/ --- 📍 Location: https://guitars4rl-map.blogspot.com/ 🔗 LinkedIn: - https://www.linkedin.com/company/the-sound-travels/ - https://www.linkedin.com/groups/9811786/ Enjoyed the video? Like, comment & subscribe for more on music, sound, and tech. 🔔 #TheSoundTravels