Theses and Student Jobs

General

  • Please contact Prof. Dr. Nils Peters for possibilities regarding theses, research internships, and student jobs.

Specific Offers

MIMO Room Impulse Response Dataset

  • Status: Filled
    Details
    • Type: Research internship, Student job
    • Description: The goals of this project are:
      • Create a novel dataset of room impulse responses that are measured using a system of distributed loudspeakers and distributed microphones (Multiple-Input and Multiple-Output).
      • Analysis and visualziation of the measured room impulse responses.
    • Skills: Knowledge in room acoustics, Familiarity with acoustic measurement equipment and software, batch scripting in Matlab or Python
    • Contact: Prof. Dr. Nils Peters

    Development of a Signal Processing Framework for Distributed Microphone Networks

  • Status: Filled
    Details
    • Type: Research internship, MSc thesis
    • Description: Have you ever wondered how microphones from all your electronic devices can be combined to create a distributed microphone array? During this project you will implement audio processing algorithms for distributed microphone arrays. These algorithms are based on recent literature (e.g., DOI:10.23919/Eusipco47968.2020.9287852) and will contribute to a modular signal processing framefork optimized for distributed microphone arrays.
    • Skills: Experience with software development and DSP implementations (preferably Python or RUST); proficiency in linear algebra and convex optimization
    • Contact: Prof. Dr. Nils Peters

Scene Representation in Wireless Acoustic Sensor Networks

  • Status: Filled
    Details
    • Type: MSc thesis
    • Description: Wireless Acoustic Sensor Networks (WASN) are present everywhere in daily life. Examples (ranging from large to small) are smart cities, home automation, cars or mobile phone ad-hoc networks. Typical applications are hearing aids, hands-free telephony, acoustical monitoring and ambient intelligence. Flexibility is often a key challenge in those topologies and require novel algorithm designed for use-cases where the nodes can (i) join and leave anytime (ii) only communicate with a subset of neighbors. Algorithms working in such scenarios synchronize locally solved problems in a global context. This can be done by splitting variables into local and global variants. Alternating local optimization and global synchronization is at heart of the Alternating Direction Method of Multipliers (ADMM) method. This work applies ADMM to Non-Negative Matrix Factorization (NMF), a parametrized representation of the acoustical scene. NMF represents the spectrogram with a codebook, overlapping codes at specific time with an activation matrix. In WASN the codebook vectors represent the acoustical scene globally, whereas the activations are indicating what is happening at a specific nodes. Once the activations are exchanged, similarity scores can give a clue about the topology and allows reconstruction of the sounds from different clusters. The goal is to derive an algorithm for the NMF variant. The implementation should be compared in terms of adaptivity, convergence and bandwidth requirements to a single node and fusion center.
    • Skills: Audio Signal Processing, Statistical Signal Processing, Convex Optimization
    • Contact: Prof. Dr. Nils Peters

Development and Evaluation of a Rotating Microphone

  • Status: Filled
    Details
    • Type: Research internship, student job, MSc thesis
    • Description: The goals is to develop a microphone apparatus that rotates with a certain velocity on a fixed circular trajectory. The accurate estimation of the instantaneous microphone position is crucial. A functional prototype will be part of future research projects.
    • Skills: Experience and interest in mechanics, electronics, and python software development.
    • Contact: Prof. Dr. Nils Peters

Learnable Acoustical Frontend for Noise Suppression

  • Status: Filled
    Details
    • Type: Research internship, MSc thesis
    • Description: In the area of neural processing for audio signals, very often a fixed acoustical frontend extracts features (e.g., bark-scaled spectrograms, MFCCs, or wavelet coefficients). Replacing such static feature extraction process with trainable equivalents let you gain an insight into their working and can produce interesting results. Your task will be finding fixed parameters which can be updated during training. The goal of your evaluation will be noise suppression. RNNoise flavored architecture gives you a starting point where you can make experiments with the acoustical frontend.
    • Skills: some experience in machine learning, background in noise suppression, some knowledge of wavelets is beneficial
    • Contact: Prof. Dr. Nils Peters

A Study on Affordable Neural Networks

  • Status: Filled
    Details
    • Type: Research internship, MSc thesis
    • Description: In the last decade neural approaches delivered top performing models in many audio fields. Two prominent examples are keyword detection in real environments and scene classification. A main objective for model design is complexity. A more regularized model gives better generalization ability and avoids very complex answers which may turn out to be wrong during testing. The focus of this work are regularization techniques reducing computational complexity as well, making those models more feasible for embedded devices. Already existing techniques are the LassoNet, top-k ReLU, fixed sparsity, matrix factorization, weight quantization and mixture of experts. You will be given two baseline models for keyword detection and scene classification. Together with your supervisor you will develop an overview and make a comparative analysis of mentioned techniques. In a second step the study should compare the trade-off between complexity and performance for selected methods.
    • Skills: experience in machine learning, interest to work with Tensorflow lite/micro or Pytorch Mobile
    • Contact: Prof. Dr. Nils Peters