Credit of the orginal paper and research is to the following:

Sharath Adavanne, Archontis Politis, Joonas Nikunen, and Tuomas Virtanen, "Sound event localization and detection of overlapping sources using convolutional recurrent neural network" in IEEE Journal of Selected Topics in Signal Processing (JSTSP 2018)

Sharath Adavanne, Archontis Politis and Tuomas Virtanen, "Localization, detection, and tracking of multiple moving sources using convolutional recurrent neural network" submitted in IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA 2019)

Refer to the original paper Git for further inormation and to download any released datasets: https://github.com/sharathadavanne/seld-net

A project of reproduction and enhancement of a DNN architecture, from the course Deep Learning by Dr. Raja Giryes, TAU

This project was conducted by Electrical Engineering M.Sc. students from Tel-Aviv university, and includes the enhancment of the SELDnet architecture for detection and localization of sound events.

This enhancement was done using a Self-Attention mechanism, as well as a exploiting the Transfer-Learning concept. Furthermore, datasets from the original paper were simply shuffeled to allow a better generelization for further predictions on any unseen data.

The results show an improvement in all evaluation metrics, and with better visualization method gives the user the ability to easily understand and evaluate the results.

Eventually, the final architecutre is presnted below: