Sensor-Controlled Prosody : Detecting Emphasis in Synthetic Speech

This study aims to investigate how a certain type of synthetic verbal communication works on both the communicator’s and the receiver’s end, specifically if the mood with which a certain sentence is expressed using a sensor can be interpreted correctly, and if so to which degree. The results in this study are based on the user’s experiences and answers, during and after testing. This study does not aim to answer why the communication works the way it does, but rather how it is experienced by the communicators. The investigation is structured around two different experiments. In the first, a group of participants use a joystick to control the melody of a monotone synthetic voice while expressing a set of sentences, each intended to convey a specific mood. The second experiment examines how these performances are interpreted by a different group of participants. The results showed that the second test group were relatively unified in their interpretations of the phrases produced by the first test group and that the interpretations were likely made deliberately, rather than being randomly guessed. In short, the moods were to a certain degree expressed and interpreted successfully.