Signals, media, machine learning — made tangible.
We are the Applied Media Systems (AMS) group at the Institute for Media Technology, Technische Universität Ilmenau. Our work connects signal processing, audio and multimedia systems, machine learning, open-source software, and hands-on teaching.
Research, software and teaching belong together.
Our public repositories turn research ideas and lecture material into executable examples: notebooks, audio-processing tools, demonstrators, optimization code, and teaching resources that can be inspected, run, changed, and reused.
Research
Audio coding, perceptual processing, filter banks, source separation, machine learning for audio, neural signal processing, room acoustics, and black-box optimization.
Open software
Python implementations, Jupyter notebooks, reproducible experiments, tutorial repositories, and research demonstrators across the AMS GitHub organization.
Teaching
Courses link theory with books, code, notebooks, interactive tutors, Moodle, seminars, and programming projects.
From classical signal processing to learned media systems.
A selection of themes represented in current and foundational AMS work.
Audio coding & psychoacoustics
Perceptual models, low-delay coding, adaptive processing, differentiable perceptual loss functions, and practical audio codecs.
Filter banks & multirate DSP
Perfect-reconstruction systems, polyphase structures, cosine-modulated filter banks, low-delay transforms, and efficient implementations.
Source separation
Time-domain and neural approaches to separating speech and music, including low-latency multichannel systems.
Machine learning for audio
Deep and recurrent neural networks, representation learning, timbre transfer, audio enhancement, and neural signal-processing models.
Black-box optimization
Derivative-free and zeroth-order methods such as Random Directions for signal-processing and machine-learning problems.
Neural room acoustics
Learning energy-decay curves and room impulse responses from room geometry, material properties, and source/receiver configurations.
Learn by running, modifying and explaining.
Many AMS courses connect lecture material directly to executable examples and interactive support.
Signal processing
Digital Signal Processing for Media Technology, Advanced DSP, Multirate Signal Processing, and Audio Coding.
Media systems
Video Coding, Multimedia Programming, and Computer Animation.
Machine learning
Machine Learning for Audio Signals with notebooks, practical examples, and research connections.
New to AMS?
Our onboarding repository is the practical entry point for new teaching and research staff. It connects TU infrastructure, teaching preparation, Moodle, GitHub, Jupyter/Colab, research orientation, publications, software, and lecture/research chatbots.
Main principle: learn the group by using its tools, teaching material, software, and research.
Open the onboarding repositoryTU account, Moodle, Nextcloud, GitHub, Jupyter, chatbots.
Courses, books, exercises, notebooks, Moodle quizzes.
Key papers, software repositories, research experiments.
Prepare and conduct a mock seminar.
Good places to start exploring.
This is a small cross-section; the AMS organization contains many more teaching and research repositories.
AppliedMediaSystems_Onboarding
Group overview, onboarding guide, teaching links, publications, research projects and chatbots.
MRSP_Tutorials
Multirate Signal Processing tutorials and notebooks accompanying the group’s teaching and book material.
AudioCoding_Tutorials
Practical examples for audio coding, perceptual processing, filter banks and related concepts.
Python-Audio-Coder
A practical Python audio coder connecting textbook concepts to executable implementations.
PsychoacousticLoss
Differentiable perceptual loss for audio and music quality and neural-network optimization.
BlackBoxOptimizerSPcomparison
Black-box and zeroth-order optimization experiments for signal processing and machine learning.
LowDelayMultichannelSourceSeparation
Research code for low-delay time-domain multichannel source separation.
timbre-transfer
Instrumental timbre transfer using disentangled representations of timbre and pitch.