Automatic Event Detection in Massive Multi-Channel Audio Signals on a Graphics Processing Unit (GPU)

Project Type: Student Project, Master/Diploma Thesis
Project Status: Open

Short description

We have data that are audio recordings that involve >1000 channels for enviromental monitoring over many hours. This leads to several Terabytes of data. Processing this large amount of data poses a significant challenge when using only a CPU. In contrast, GPU are optimized for highly parallel processing. The aim of this project/thesis is to develop a framework that allows to process those audio channels in real-time with a latency of  up to a few seconds. 

The current choice of programming language is Python and/or Julia. This is still negotiable, but the choices should not be restricting due to licence issues, as is the case with Matlab.

Your Tasks

  • Review of toolboxes and approaches available 
  • Implementation of a framework in Python and/or Julia
  • Implementation of basic processing algorithms
  • Documentation

Your Profile/Prerequisites

  • Motivation and interest in the topic
  • Background in Signal Processing
  • Experience with either Matlab, Python or Julia
  • Only in teams of 2-4 students

Contact:

 Martin Hagmüller (hagmueller@tugraz.at or 0316/873 4377)