GitHub Activity

Band Project

Solo Project

Project thesis

List of selected projects

pyDCM (python3)

for FEV GmbH (2013)

 

The DCM file format is the standard for engine calibration and it contains the parameters. A python library was written such that it enabled processing of the DCM files in python3 and allowed operations like read, write, copy, delete, compare, etc.

 

 

 

Simulation of calibration corrections (VBA, later python3)

for FEV GmbH (2014-2015)

 

A powerful tool capable of loading DCM files, using real drive test files and simulating the changes in engine and emission performance was developed. This enabled the calibration engineer to make first assumptions without the need of driving the same test after every change in the calibration.

 

 

 

Prediction of water condensation in exhaust gas recirculation (python3)

for FEV GmbH (2015 - 2016)

 

One of the biggest challenges in diesel engine calibration is the condensation of water in the exhaust path, which can lead to engine failure. A tool was developed to predict the chemical composition at several spatial locations within the exhaust path, and to find the correct engine calibration for avoiding the condensation of water.

 

 

 

Raspberry Pi Audio Streaming (bash, icecast2)

for Hotel Europa (2017)

 

As an inexpensive solution to have the same music playing wirelessly on two different levels in the club “Hotel Europa” in Aachen, two Raspberry Pi were modified. The first Pi was set as an icecast2 streaming server, while the second streamed from it. Bash scripts were used  to start the stream automatically after powering the devices on, and thereafter, no further setup was needed.

 

 

 

MIDI Channel reassignment (MAX 8)

for MODUS (2018 - ongoing)

 

MIDI controllers sending the same CC/Notes and MIDI-Channel cannot be used separately on MIDI mapped controls in Ableton Live 9. To separate these, the MIDI-Channel of each controller was changed to a unique MIDI-Channel and then forwarded to Ableton and vice versa. This enabled the use of 10+ different MIDI controllers without the need of changing settings on the controller itself.

 

 

 

Automatic Detection of Acoustic Signals in Traffic (python3, Tensorflow/Keras, SQLite)

for VKA Institute at RWTH-Aachen (April 2019 - April 2020)

 

With the rise of autonomous driving cars and acoustically isolated interior in cars, detecting acoustical signals in traffic has become a necessity for safety. A pythonic software framework was developed, which included functions like:

• Import/export of training data
• Annotation of training data
• Dynamic selection of settings/parameter to train neural networks
• Batch training of neural networks
• Automated documentation of settings and results
• Automated testing and ranking
• Visualisation of test results
• The final neural network was capable of detecting and distinguishing acoustic signals with accuracy greater than 90%.

 

(Download PDF)

 

 

 

Quantised Audio Streamer (Max4Live)

for everyone (April 2020 - ongoing)

 

This Max4Live effect enables users to “jam” over the internet in Ableton Live. Because of the inevitable latency due to sending lossless audio, the playback was programmed to get delayed for a selected amount of bars. Within the delayed time, the audio is saved and sent in packets to the other user. The received files are then played back in a quantised manner, and therefore, perfectly in sync.

 

(https://github.com/bakaj/MODUS-Streamer)

Contact