We offer two 6-month project groups

• Nachrichtentechnik / Communications Engineering (L.048.10809 / L.048.98005)
• Deep Learning (L.048.28049, L.048.98049)

and one 12-month project group

• Communications and Speech Processing (L.048.28501, L.048.98501)

Topics change from semester to semester.

In the following we describe some typical topics to give an idea of what the work will be concerned with. But, please bear in mind that the actual concrete topic may be different and will be announced in a meeting in the beginning of the semester. You can find the meeting on our homepage under "Lehre / Teaching".

Acoustic sensor networks are a new and exciting research topic. Processing non-synchronously sampled multi-channel signals is a challenging task. The current project group will focus on methods for processing signals from spatially distributed acoustic sensor nodes.

In previous project groups we have sucessfully developed sensor nodes based on Rasperry Pi's Model 3b+. This tiny but powerfull device are connected via an AD-HOC network and form together our signal processing platform. Futhermore we have setup a software framework called MARVELO which eases the process of task and algorithm distribution.

The promoted project group can focus on implementing algorithms, e.g.,  for localizing  speakers, enhancing signals and identifiying acoustic events, just to name  a few possible tasks. The requirements are as follows:

• You shoud be able to read and write Python Code.
• You have an idea what discrete signals are and how to process them in a block-oriented fashion

You have further questions? Connact us via Email or visit us at our offices any time. The responsible contact person you find on the right side.

Project Outline

• Phase 0: Project coordination meeting, task assignement
• Phase 1: Selecting algorithms and paper reading
• Phase 2: Implementation of first prototypes
• Phase 3: Online testing on Raspberry PI network
• Phase 4: Bugfixing & Optimization
• Phase 5: Real-World tests
• Phase 6: Documentation & Presentation

During this project we tackle different topics in speech and audio processing with current maschine learning techniques. As the result of the project a short demonstration should be created, which may be a live demo, a homepage or other ways of bringing your subject to life.

Possible subjects are:

• Automatic speech recognition
• Acoustic event detection
• Anomaly detection
• Blind source separation

Appointment for first meeting is published on our "Teaching" website.