Scientists have created an algorithm to assess the emotional coloring of speech
Specialists Massachusetts Institute of technology, laboratory Artificial Intelligence Laboratory and Institute of Medical Engineering and Science developed an algorithm that can analyze the emotional coloring of speech. What most people perceive not even thinking to electronics is extremely challenging.
Using the mechanisms of deep learning, the system is able to analyze the audio and even text transcription, determining the overall tone of speech with an accuracy of 83%.
The study participants wore a device which the developers call Samsung Simband. Judging from the appearance, this smart watch Galaxy Gear S. In this case, they were selected due to the presence of a variety of sensors, which helped to gather more data during the conversation. It should be noted that the assessment was not made directly on the wearable device in real time. The speech was recorded, then translated to text and then the system gave all these data, complemented by information from the sensors hours.
The developers of the algorithm talking about the benefits of such technologies for people with Asperger’s syndrome are prone to anxiety States and other mental disorders, which interfere with correctly perceiving the emotional coloring of the speech, and therefore often its meaning. But this same technology can be used for other purposes that are not related to neurological disorders.