Department of
Computational Perception
|
Department of
Computational Perception
|
|
Strategic FExFE Project: Deep Learning
Full Project Title: Strategic FExFE Project on Deep Learning
Sponsor: Austrian Research Promotion Agency (Österreichische Forschungsförderungsgesellschaft; FFG)
Duration: 44 months (May 2015 - Dec. 2018)
Persons involved at our institute:
Project Partners
Short Description
This is a strategic cooperation, in the intersection of basic and applied research, between FFG-supported COMET Center SCCH and the Dept. of Computational Perception (CP) of JKU Linz, with the goal of performing joint research on questions of common interest in the field of Deep Learning. It currently funds one full-time Ph.D. student at CP.
Publications
Aktuelle Entwicklungen in der Automatischen Musikverfolgung.
Matthias Dorfer und Andreas Arzt.
Workshop: Musik trifft Informatik, 2017.
A Hybrid Approach to Acoustic Scene Classification Based on Multi-channel I-Vectors and Convolutional Neural Networks.
Hamid Eghbal-zadeh, Bernhard Lehner, Matthias Dorfer, and Gerhard Widmer.
In Proceedings of the 25th European Signal Processing Conference (EUSIPCO), 2017.
Drum Transcription from Polyphonic Music with Recurrent Neural Networks.
Richard Vogl, Matthias Dorfer, and Peter Knees.
In Proceedings of 42nd IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2017.
Towards Score Following in Sheet Music Images. (PDF)
Matthias Dorfer, Andreas Arzt, and Gerhard Widmer.
In Proceedings of 17th International Society for Music Information Retrieval Conference (ISMIR), 2016.
Recurrent Neural Networks for Drum Transcription.
Richard Vogl, Matthias Dorfer and Peter Knees.
In Proceedings of 17th International Society for Music Information Retrieval Conference (ISMIR), 2016.
On the Potential of Simple Framewise Approaches to Piano Transcription.
Rainer Kelz, Matthias Dorfer, Filip Korzeniowski, Sebastian Böck, Andreas Arzt and Gerhard Widmer.
In Proceedings of 17th International Society for Music Information Retrieval Conference (ISMIR), 2016.
Downbeat Estimation from Beat Synchronous Features with Recurrent Neural Networks.
Florian Krebs, Sebastian Böck, Matthias Dorfer and Gerhard Widmer.
In Proceedings of 17th International Society for Music Information Retrieval Conference (ISMIR), 2016.
A Cosine-Distance based Neural Network for Music Artist Recognition using Raw I-vector Features.
Hamid Eghbal-zadeh, Matthias Dorfer and Gerhard Widmer.
In Proceedings of the 19th International Conference on Digital Audio Effects (DAFx16), 2016.
Deep Linear Discriminant Analysis. (PDF, CODE)
Matthias Dorfer, Rainer Kelz, and Gerhard Widmer.
In Proceedings of the International Conference on Learning Representations (ICLR), 2016.
CP-JKU Submission for DCASE-2016: A Hybrid Approach Using Binaural I-Vectors and Deep Convolutional Neural Networks.
Hamid Eghbal-zadeh, Bernhard Lehner, Matthias Dorfer and Gerhard Widmer.
1st and 2nd place winners