Exemplary embodiments include an assistive device to facilitate social interactions in autistic individuals by identifying emotions using a voice-detecting machine learning algorithm that extracts emotion content from an audio sample input and outputs the emotional content to a user through a device. This device may be a portable, concealable, real-time and automatic device that may receive and process an audio input. The audio input may be analyzed using a machine learning algorithm. The device may output the closest emotional match to the autistic user. The output may be tactile in nature such as a vibration pattern that is different for different identified emotions.