Spanish-MEACorpus-2023: A Multimodal Spanish Corpus for Emotion Analysis from YouTube

Spanish-MEACorpus-2023 is a multimodal Spanish corpus for emotion analysis constructed from YouTube videos. The dataset includes segmented speech excerpts annotated with emotion labels and speaker gender, along with temporal boundaries and manually revised transcriptions. Each instance is linked to a YouTube video identifier, enabling reproducibility and rehydration of the original multimedia content. Due to copyright restrictions, audio and video files are not redistributed. Instead, the dataset provides identifiers, timestamps, and transcriptions, together with code to automatically download and segment the original videos from YouTube. The dataset was introduced in the associated scientific publication and is made available here following the FAIR principles to ensure long-term reuse.