Neuromorphic Acoustic Detection and Recognition

Chr. Kehling TU Ilmenau

Goals

The NADAR project aims to use neuromorphic microsensors to develop an energy-efficient, robust, and resource-saving audio system for the detection, recognition, and classification of acoustic anomalies in industrial environments.

The human auditory system served as inspiration and combines bio-inspired sensor technology with artificial neural networks.

Some of the main aspects of the project are:

  1. Energy-efficient acoustic monitoring
  2. Localization of the source of the acoustic event in a complex and changing acoustic environment
  3. Further processing to classify and detect only the relevant data for signal analysis
  4. Adapting signal processing and tuning sensor characteristics as efficiently as possible to respond to changes in the environment or in the emitted sounds
Chr. Kehling TU Ilmenau

Milestones

  1. Hardware Exploration: Investigation of possible combinations of cantilever sensor arrays and their optimal parameter settings.
  2. Software Integration: Integration of state-of-the-art bio-inspired neural technologies for automated acoustic scene recognition.
  3. Application Testing: Creation and evaluation of various complex listening scenes with multiple sound sources for the detection and localization of sound events.
  4. Energy Consumption Optimization: Optimizing system parameters with regard to sustainability and minimal computational load.
     

Project Details

Duration: July 1, 2024, through December 31, 2027
Funded by the Carl Zeiss Foundation

Project Team

Dipl.-Ing. Christian Kehling, Dr.-Ing. Stephan Werner, Dr. rer. nat. Tzvetan Ivanov

 

NADAR is a project of the Ilmenau School of Green Electronics (ISGE), a central, interdisciplinary educational and research organization.