Go to JKU Homepage
Institute of Computational Perception
What's that?

Institutes, schools, other departments, and programs create their own web content and menus.

To help you better navigate the site, see here where you are at the moment.

Detail

Three Awards for CP-Researchers at DCASE Workshop 2024
 

We’re happy to announce that Jonathan Greif received the Best Student Paper Award for his paper  “Improving Query-by-Vocal Imitation with Contrastive Learning and Audio Pre-Training” at the DCASE Workshop 2024. Additionally, our research team received two Judges’ Awards for their submissions to the “Sound Event Detection” and “Language-Based Audio Retrieval” tasks of the IEEE AASP DCASE Challenge 2024.

award

The Workshop on Detection and Classification of Acoustic Scenes and Events (DCASE), held in Tokyo from October 23–25, 2024, is a key international forum for advancements in audio signal processing. Accompanied by the IEEE AASP DCASE Challenge, opens an external URL in a new window, the event brings together researchers to present and discuss the latest developments in the field.

Our researchers Paul Primus, Florian Schmid, Jonathan Greif, Tobias Morocutti, and Gerhard Widmer received three awards at this year’s workshop, recognizing their contributions to the field of intelligent acoustic signal processing. 

Best Student Paper Award for Vocal Imitation Research

The paper “Improving Query-by-Vocal Imitation with Contrastive Learning and Audio Pretraining, opens an external URL in a new window” was recognized with the Best Student Paper Award. This work explores how contrastive learning and pre-trained neural networks can enhance Query-by-Vocal Imitation (QBV) systems, which allow users to search for sounds by mimicking them vocally.

Judges’ Awards for Challenge Contributions

Two submissions from CP researchers received Judges’ Awards in the DCASE 2024 Challenge:

  1. Sound Event Detection (Task 4):
    The work “Improving Audio Spectrogram Transformers for Sound Event Detection through Multi-Stage Training, opens an external URL in a new window” introduces a training routine designed to address the scarcity of high-quality sound event detection datasets. This approach achieved the top rank in Task 4, opens an external URL in a new window.
  2. Language-Based Audio Retrieval (Task 8):
    The work “A Knowledge Distillation Approach to Improving Language-Based Audio Retrieval Models, opens an external URL in a new window” presents a novel method that leverages estimated audio-caption correspondences during contrastive language-audio training. A system based on this method achieved the top rank in Task 8, opens an external URL in a new window.

You can learn more about the DCASE community on their official website, opens an external URL in a new window.