STAR: Speech Training, Assessment and Remediation

The STAR system is a prototype speech assessment and training tool that uses an animated Star character to interact with and encourage children in speech learning and production tasks. These tasks are incorporated into computer games. The STAR character acts as a guide and motivator throughout the games. The system, when completed and functioning, will start with basic phoneme recognition tasks, and progress through phoneme production tasks, first in isolation, then in simple and eventually more complex phonetic contexts, and finally in sentences.

Currently we are in the process of evaluating one of the games. The game is a simple speech production game in which a child is asked to produce words drawn from a minimal pair that contrast a phoneme to be trained with one the child has already achieved. For example, if a child produces /w/ correctly, but also substitutes /w/ for /r/, we would use the minimal pair Rhonda and Wanda as names of characters in the game. The game proceeds by asking the child to say one of the words, captures and analyzes the child's utterance to determine how closely it matches recognition models for both contrasting words, and animates the character whose name most closely matches the child's utterance. Feedback regarding the accuracy of the child's production is given by spoken response from the star character, and from a meter which points to the character whose name was spoken. The game is being evaluated for its ability to effectively decrease the length of acquisition time for the sounds being trained.

The STAR project is required to recognize and rate children's utterances. Because so little work has been done with children's speech, this is an area of research in our lab. Other areas of research include assessing and quantifying the level of difficulty of pronunciation of utterances, and the overall effectiveness of different rates of advancement throughout the game.

In its present form, the STAR system allows us to collect data on children's speech production, as well as information about how children interact with a believable computer agent in this domain, and thus provide information needed to progress toward our research goals. Long range goal for the STAR project is to combine the latest technologies involving believable animated agents, speech recognition, artificial intelligence, and speech synthesis to create a speech learning tool that children find enjoyable and easy to work with. More immediately, the speech analysis and recognition components being developed will provide clinicians supervising speech therapy with information about a child's speech production that they could not determine through listening alone, thus allowing them to more accurately correct and work with the child. We expect that, by making the games simple and enjoyable, children will be more motivated to work on their speech production tasks.

Projects | Publications | Related Links | Staff |
Facilities | Events | Positions