This paper presents the EmoSPeech 2024 shared task, which was organized in the IberLEF 2024 workshop within the framework of the 40th International Conference of the Spanish Society for Natural Language Processing (SEPLN 2024). The objective of this shared task is to study the field of Automatic Emotion Recognition (AER), which is becoming increasingly important due to its impact on various fields, such as healthcare, psychology, social sciences, and marketing. Specifically, two tasks are proposed and evaluated separately. The first task deals with AER from text, which focusing on feature extraction and identifying the most representative feature of each emotion in a dataset created from real-life situations. The second task deals with AER from a multimodal perspective, which requires the construction of a more complex architecture to solve this classification problem. The ranking includes the results of 13 different teams, each of which proposed a novel approach to the problem.