Unlocking the World of Sounds: Machines that Hear and Feel

Imagine starting your morning with birdsong outside your window, a sound that effortlessly lifts your mood. Now, imagine the harsh clanging of construction machinery—it’s not just annoying, it actively changes your emotional state. This simple yet profound interplay between sound and emotion inspired Yuanbo Hou’s doctoral research, titled "Advancing Machine Listening: Understanding Acoustic Scenes, Events, and the Emotions They Evoke."

Hou’s groundbreaking work explores how artificial intelligence can not only detect and classify sounds but also understand their emotional impact. His research promises smarter cities, better health monitoring, and even AI tools that can empathize with our acoustic environments.

Making Machines Listen Better

At its core, Hou’s research pushes beyond traditional sound recognition systems, which focus solely on identifying audio events like barking dogs or honking horns. His work introduces a three-dimensional model encompassing acoustic scenes, audio events, and the emotions they evoke.

"Understanding how sounds make people feel isn’t just a technical challenge," says Hou. "It’s about making machines that truly connect with human experiences."

From Events to Emotions: The Science of Soundscapes

Soundscapes are rich tapestries of auditory events, from serene parks to bustling streets. Hou’s research takes inspiration from soundscape studies, which historically relied on labor-intensive surveys to measure how different environments affect people emotionally. Leveraging deep learning models, Hou automates this process, offering unprecedented insights into how sounds affect human well-being.

By introducing tools like the SoundEQnet and the Large Language Model-based Soundscape Describer (LLM-SSD), his research allows machines to analyze audio, connect it to emotional responses, and even generate descriptive captions. This makes soundscape research faster, smarter, and more accessible.

Empathy in AI: A Future of Intelligent Listening

The implications of this research stretch across disciplines. Picture urban planners designing quieter cities based on AI-analyzed soundscapes or healthcare tools that monitor stress levels through audio cues. The possibilities extend to entertainment, virtual assistants, and even emergency response systems, which could adapt based on the emotional tone of an environment.

“By modeling emotions alongside sound, we’re teaching machines to understand the human world better,” Hou explains.

Toward a Harmonious Future

Hou’s work represents a significant leap forward in machine listening, blending technical innovation with a profound understanding of human emotion. It’s not just about recognizing sounds—it’s about creating technologies that resonate with the rhythms of our daily lives.

As we envision cities, homes, and devices that listen and respond thoughtfully, Hou’s research offers a glimpse of a future where machines don’t just hear—they truly understand.

Read a more detailed summary or the entire PhD

Chinese Government Award

Dr. Yuanbo Hou has been awarded the “2023 Chinese Government Award for Outstanding Self-financed Students Abroad” for his work during his PhD study at Ghent University.

This award, starting in 2003, was founded by the Chinese government with the purpose of rewarding the academic excellence of self-financed Chinese students studying overseas. Only those with outstanding performance in their PhD studies or Post-doctoral research will be considered by the award selection panel and no more than 650 young talents will be granted the award each year all over the world.

-

PhD Title: Advancing Machine Listening: Understanding Acoustic Scenes and Events and the Emotions They Evoke

Promotors: Dick Botteldooren en Sarah Verhulst

-

Editor: Jeroen Ongenae - Illustrator: Roger Van Hecke