The Ryerson Audio-Visual Database of Emotional Speech and Song (RAVDESS) is a collection of recordings made from 24 actors (12 male, 12 female) speaking and singing with various emotions, in a North-American English accent. The entire database contains 7,356 high-quality video recordings of emotionally-neutral text, spoken and sung with a range of emotions. The speech set consists of the 8 emotional expressions: neutral, calm, happy, sad, angry, fearful, surprise, and disgust. The song set consists of the 6 emotional expressions: neutral, calm, happy, sad, angry, and fearful. All emotions except neutral are expressed at two levels of emotional intensity: normal and strong. Audio, visual and audio-visual exemplars from the database have been validated in a perceptual experiment involving 297 participants. Detailed acoustical analyses have also been conducted. Overall, speech and song convey emotions using similar acoustic cues but song samples longer in duration, with higher pitch floor and louder levels. The RAVDESS is freely available for scientific-research and general use under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License