Odyssey 2024 - Speech Emotion Recognition Challenge: Dataset, Baseline Framework, and Results
Lucas Goncalves, Ali N. Salman, Abinay R. Naini, Laureano Moro Velazquez, Thomas Thebaud, Leibny Paola Garcia, Najim Dehak, Berrak Sisman, Carlos Busso
Code Available — Be the first to reproduce this paper.
ReproduceCode
Abstract
The Odyssey 2024 Speech Emotion Recognition (SER) Challenge aims to enhance innovation in recognizing emotions from spontaneous speech, moving beyond traditional datasets derived from acted scenarios. It offers speaker-independent training, development, and an exclusive test set, all annotated for the two tracks explored in this challenge: categorical and attribute SER tasks. This initiative promotes collaboration among researchers to develop SER technologies that perform accurately in real-world settings, encouraging researchers to explore innovative approaches that leverage the latest advancements in audio processing for SER. In this paper, we provide a detailed description of the baseline, leaderboard, evaluation of the results, and a discussion of the key findings. The competition website with leaderboards, links to baseline code, and instructions can be found here: https://lab-msp.com/MSP-Podcast_Competition/leaderboard.php