The voice carries a wealth of information, even if no words are spoken. From laughs to noncommital grunts, we use many nonverbal exclamations in our everyday interactions to indicate how we feel and what we intend to do. The purpose of this experiment is to learn more about how listeners interpret such emotional vocalizations.

In order to understand how information is extracted from non-speech sounds, it would be very valuable to be able to synthesize them and modify different acoustic characteristics at will. We have developed an open-source algorithm for sound synthesis at Lund University in Sweden, which we used to reproduce a number of real human vocalizations. We are now interested in finding out how these synthetic sounds are perceived - whether they sound realistic and can portray emotion.

You will hear 60 non-speech sounds. Half of these sounds were taken from YouTube videos of people engaged in emotionally charged activities, and the other half are synthetic, computer-generated (all mixed up in random order). For each sound, select ONE emotion it portrays and click the scale to indicate how certain you are. To repeat a sound, click the picture of a loudspeaker. Replay sounds as many times as you like, but don't overdo it - try to do the test as quickly as possible!

Find a quiet spot to do this experiment. It takes at most 10 minutes, but you can also stop earlier, the data is saved automatically. If at some point the sound disappears, please wait a few seconds - it might be a bad internet connection or a server error. Your participation is completely anonymous, and you will not be asked to provide any personal information.

Put on headphones and make sure the volume is not too high. Ready? OK then, click "Begin"!