Fabian Ostermann
| Room: | OH14 / R234 |
| Phone: | +49 231 755-7742 |
| E-Mail: | fabian.ostermann@tu-dortmund.de |
| ORCiD: | 0000-0002-8365-3634 |
| Github: | @FabianOstermann |
Research interests (Forschungsinteressen)
- AI Computer Music and Algorithmic Music Composition
- (Generative) Reinforcement Learning
- Evolutionary Algorithms (Neuroevolution)
- Procedural Content Generation (for Games)
Teaching (Lehre)
Lehrveranstaltungen
- SS24 Fachprojekt Musikinformatik
- WS23/24 Fachprojekt Digital Entertainment Technologies
- SS23 Übung zur Vorlesung Praktische Optimierung
- WS22/23 Vorlesungseinheit in der Ringvorlesung Musikdatenanalyse zum Thema Automatische Komposition
- WS22/23 Fachprojekt Digital Entertainment Technologies
- SS22 Fachprojekt Digital Entertainment Technologies
- WS21/22 Praktikum zur Vorlesung Einführung in die Progammierung
- WS20/21 Übungseinheiten in der Ringvorlesung Musikdatenanalyse
- SS2017 Übungseinheiten in der Ringvorlesung Musikdatenanalyse
Betreute Abschlussarbeiten
- “Generierung von Orchesterstücken mit Transformer-Modellen”
- “Transformer based drum track generation for electronic dance music”
- “Einsatz von Adversarial Reinforcement Learning zur automatischen Erstellung spielbarer Level in einem einfachen Rennspiel”
- “An Evolutionary Approach to Recreating Vector Graphics from Raster Graphics”
- “Hyperparameteroptimierung eines neuroevolutiven Ansatzes für ein Autorennspiel”
- “Grafische Darstellung hochdimensionaler Fuzzy Cluster durch Projektion und Alpha-Schnitte”
Publications
Articles (peer reviewed)
- Adaptation and Optimization of AugmentedNet for Roman Numeral Analysis Applied to Audio Signals (PDF)
Leonard Fricke, Mark Gotham, Fabian Ostermann, Igor Vatolkin
International Conference on Evolutionary and Biologically Inspired Music and Art (EvoMUSART), Springer, 2024, pp. 146–161.
- 🏆 Adaptive video game music as a multi-objective benchmark for conditional autoregressive models (PDF)
Fabian Ostermann
Proceedings 33rd Workshop Computational Intelligence, KIT Scientific Publishing, 2023, pp. 199–214
- Application of Neural Architecture Search to Instrument Recognition in Polyphonic Audio (PDF)
Leonard Fricke, Igor Vatolkin, Fabian Ostermann
International Conference on Evolutionary and Biologically Inspired Music and Art (EvoMUSART), Springer, 2023, pp. 117–131
- Musical Genre Recognition based on Deep Descriptors of Harmony, Instrumentation, and Segments (PDF)
Igor Vatolkin, Mark Gotham, Néstor Nápoles López, Fabian Ostermann
International Conference on Evolutionary and Biologically Inspired Music and Art (EvoMUSART), Springer, 2023, pp. 413–427
- AAM: A Dataset of Artificial Audio Multitracks for Diverse Music Information Retrieval Tasks (PDF)
Fabian Ostermann, Igor Vatolkin, Martin Ebeling
EURASIP Journal on Audio, Speech, and Music Processing (2023)
- Artificial Music Producer: Filtering Music Compositions by Artificial Taste (PDF, VIDEO)
Fabian Ostermann, Igor Vatolkin, Günter Rudolph
Proceedings of the Conference on AI Music Creativity (AIMC), 2022
- Evaluating creativity in automatic reactive accompaniment of jazz improvisation (PDF)
Fabian Ostermann, Igor Vatolkin, Günter Rudolph
Transactions of the International Society for Music Information Retrieval (TISMIR) 4 (2021), Nr. 1
- An evolutionary multi-objective feature selection approach for detecting music segment boundaries of specific types (PDF)
Igor Vatolkin, Fabian Ostermann, Meinard Müller
Proceedings of the Genetic and Evolutionary Computation Conference (GECCO), 2021, pp. 1061–1069
- Evaluation rules for evolutionary generation of drum patterns in jazz solos (PDF)
Fabian Ostermann, Igor Vatolkin, Günter Rudolph
International Conference on Evolutionary and Biologically Inspired Music and Art (EvoMUSART), Springer, 2017, pp. 246–261
Theses
- Modellierung des menschlichen Musikgeschmacks mit neuronalen Netzen
Fabian Ostermann, Master-Thesis, TU Dortmund, 2021
- Ortung einer akustischen Quelle im Roboterfußball mittels Multilateration
Fabian Ostermann, Studienarbeit, TU Dortmund, 2020
- Erzeugung perkussiver Muster zur Echtzeitbegleitung von Jazz-Soli
Fabian Ostermann, Bachelor-Thesis, TU Dortmund, 2015
Project Pages
- Dataset of Artificial Audio Multitracks (AAM)
The dataset contains 3,000 artificial music audio tracks with rich annotations. It is based on real instrument samples and generated by algorithmic composition with respect to music theory.
- ArtificialSongGenerator (Github)
Algorithmic music composition tool used to compile the AAM dataset.
- EAR Drummer (Github) (YouTube demo, more audio demos)
EAR Drummer is a interactive music system that generates rhythmical drum patterns and bass lines. It produces midi output that provides backing music to accompany a Jazz soloist in realtime. It uses Evolutionary Computing, is fully Autonomous and Reacts to midi input.