Introduction to the Sensory Cognition Group

Sensory cognition group of the AI-department of the University of Groningen is a research group that studies all aspects of sound perception. In particular it focuses on (sound) perception in the real world, i.e. how perception processes contribute to all the things we do when we are not part of an experiment. The group does this by a combination of theory developement, modelling, and experiments.

The group focuses on a number of topics

• Sound recognition in unconstrained acoustic environments
• Sound source recognition of real-world sounds
• The perceptual effects of room-acoustics
• The link between physics, spectro-temporal patterns, and cognition
• Models of auditory attention
• Condition monitoring of complex systems
• Cognition as necessary prerequisite for dealing with an unconstrained world
• Cognitive computational neuroscience

What is auditory cognition?

Auditory cognition is about the processes between the reception of a sound and any behavior this might lead to. For example if someone greets you, you respond, when you hear a glass breaking you look around, if you like a sound you might linger for a while, and if you do not like the the ever increasing number passing aircraft over your house you complain. The sound related processes that help you to make sense of the world around you, that allow you to recognize and like or dislike sounds, and that lead you to respond in the way you do are all part of auditory cognition.

A summary of our current research:

Signal-driven sound processing for uncontrolled environments

Modern signal processing and recognition techniques need assumptions about the signal they are goning to process, e.g., that the signal is speech or music, this assumes a role for a user who makes sure that the input is suitabe for the system. This of course entails that the system can never be fully autonomous. Link to research Dirkjan

Sound event recognition

Sounds are not just acoustic energy: it is highly structured acoustic energy. And the structure that the sound sources impose is preserved while the sounds travel to your ear. We try to estimate these structures from the sounds and use knowledge about which sound produces which structure to recognize the sounds. [link]

The role of context in sound event recognition

Often the sound of a source contains simply not enough information to recognize it correctly. Take the sound of a whistle. That it is a whistle sound is not so difficult to determine. But if you hear in the context of a noisy busy place where it is directly followed by a a sound that can be a leaving train, than you might conclude that the you heard the sound of a train attendance. The whistle enhances the probability that the sound is a train and the train helps to interpret the whistle specific type of whistle. We try to incorporate this type a knowledge in our recognition systems.

Soundscape research

Soundscape research is aimed at the design and analysis of our sonic living environment. Compared to traditional acoustic research is does not focus exclusively on acoustic aspects of sound for design and analysis, but it takes into account other important aspects social and psychological factors and the diversity of listeners and listening modes. Soundscape aspects have become more and more important because our living environments have become more and more noisy and we hear pleasant sounds not as much as many people would like to. Soundscape research addresses these and other issues.

Well-being and the sonic environment

Sounds have a surprisingly strong impact on our well-being. We all enjoy some music and hate other types (e.g. soulless commercial muzak). Some sounds, such as a loud passing car, attract attention easily. If they are irrelevant to your current activities these sounds "steal" some of your time which can become quite annoying. If these sounds persist in disturbing you against your will they raise you anxiety-level. You feel a bit more stressed, and if this situation persists your overall well-being and even your health might even be influenced. We aim to understand the positive and negative influences of sound on well-being.

Perception and attention

It is still far from clear what people actually do when they listen. Actually it quite possible that it is sensible to make a separation between hearing and listening. The hearing process is a background process that is always active and that informs us of changes or salient in the sonic environment and can force us to pay attention to certain events. The listening process is an active knowledge and attention intensive that searches in the signal for evidence of more or less expected events. For example when you expect a sentence in one language, but you hear a sentence in another (known) language you are likely to misinterpret the first few words. An fairly accessible description of the processes involved in audition is available here [Link to Audition: from sound to sounds"]

Current PhD-projects focus on
- Key word spotting
- Optimizing grouping strategies for Auditory Scene Analysis
- Auditory sensitivity of people with multiple mental disabilities