There are many complex human emotion perceptions and expressions (e.g., angry emotions affect facial expressions, voices, and language). Here is an open dataset with audio-videos tied together and emotionally labeled.
A dataset called The Ryerson Audio-Visual Database of Emotional Speech and Song (RAVDESS) shows 12 male and 12 female professional actors uttering in emotional situations given as calm, happy, sad, angry, fearful, surprise, and disgust. Recorded audio-video data. Although there is not a lot (audio-video: 2880 audio files + 2024 songs files, audio only: 1440 files audio, 1012 songs), I plan to use them in future emotional studies.