This section lists some of the key dissemination activities stemming from the project’s lifespan and beyond. This webpage will be updated periodically to reflect the most recent activities which are taking place beyond the official end of the project’s duration.
Publication related
[P.1] The following paper was accepted at the 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP-2021), which is an international, peer-reviewed, flagship conference in the field of acoustics, speech and signal processing: Petros Giannakopoulos, Aggelos Pikrakis and Yannis Cotronis, “A Deep Reinforcement Learning Approach to Audio-based Navigation in a Multi-speaker Environment”, ICASSP-2021, 6-11 June 2021, Toronto, Ontario, Canada. The work presented in the paper stems from WP4 of the project on acoustic scene analysis and it is based on a modern machine learning technique from the field of reinforcement learning to solve the problem of audio-based scene analysis. ICASSP is an annual conference with several sessions on tasks of immediate interest to the project, including audio event detection and audio scene classification.
[P.2] The following paper was accepted at the IEEE Access journal (https://ieeeaccess.ieee.org), which is a high visibility, international, peer-reviewed IEEE journal with impact factor 3.367: Petros Giannakopoulos, Aggelos Pikrakis and Yiannis Cotronis, “Improving Post-Processing of Audio Event Detectors Using Reinforcement Learning,” in IEEE Access, vol. 10, pp. 84398-84404, 2022, doi: 10.1109/ACCESS.2022.3197907. The work presented in the paper stems from WP4 of the project on acoustic scene analysis and it is based on a deep reinforcement learning technique to provide a generic solution to the problem of improved post-processing of audio event detectors, irrespective of the analysis method under consideration.
Algorithm challenge related
[AC.1] We participated in Task 4 of the DCASE-2022 algorithmic challenge ( https://dcase.community/challenge2022/task-sound-event-detection-in-domestic-environments-results) which is related to audio event detection in a home environment by submitting an algorithm based on multi-task learning and deep variational autoencoders, prepared by Petros Giannakopoulos (Ph.D. candidate and member of the project team) and Aggelos Pikrakis (PI of the project). A technical report of our participation can be accessed online at the website of the DCASE challenge (and here)
Invited talks
[IT.1] Aggelos Pikrakis,“A Deep Reinforcement Learning Approach to Audio-based Navigation and Audio Source Localization”, Institute of Nuclear and Particle Physics, NCSR “Demokritos”, Tuesday 12 Apr 2022, 16:00-17:00 (talk link) and (talk ppt, restricted access, for project review purposes only)
Industry related
[IND.1] The Soundscapes Project objectives were presented at a workshop organized in 2021 by the Greek company Libra AI Technologies, https://www.libramli.ai, Serafeio Complex, Athens, Greece.
[IND.2] Communication was established via online discussions and site visits, with a London-based English company and a Greek one to investigate whether the methods developed in the Soundscapes project can be exploited in a business context, e.g., as analysis/synthesis modules. Respective letters of intent can be downloaded here (restricted access, for project review purposes only).