VOSOME

The dataset contains features and labels of 1000 audios from real social media conversations. The platform used was WhatsApp and they are Spanish speakers. Labels_using_audios: File with the labelling of 5 expert based on audio. Labels_using_text: File with the labelling of 3 expert based on text (audio manual transcription). Speech_features: Features obtained without applying any filter. The features are the eGeMAPSv02 set of opensmile. Speech2text: Transcription of the audios. It includes a manual transcription, and automatic transcription using amazon and google API. User_questionnaire: Metadata of the subjects.