SpeakingFaces. A large-scale publicly-available visual-thermal-audio dataset designed to encourage research in the general areas of user authentication, facial recognition, speech recognition, and human-computer interaction.