Toys That Have a Voice
One of the most interesting and informative sessions at SpeechTEK 2002 was the Real Solutions Seminar presented by Dave Peterson, vice president of Technology Acquisition at Hasbro. Peterson described two toys that Hasbro released this year: Aloha Stitch and R2D2. Both toys are based on popular movie characters and both interact with a child using speech recognition.
Speech recognition has a long and spotty history in the toy industry. Radio Rex was the first commercial toy to respond to voice commands. Produced in 1922 by the Elmwood Button Co., Radio Rex predates computers by more than 20 years. Rex is a brown bulldog made of celluloid and metal that responds to its name by leaping out of its house. The dog is controlled by a spring that is held in check by an electromagnet. The electromagnet is sensitive to sound patterns containing acoustic energy around 500 Hz, such as the vowel in "Rex." The acoustic trigger interrupts the current to the electromagnet allowing the spring to propel Rex out of its house. Mike Cohen, vice president of Dialog R&D at Nuance, owns one of the few Radio Rex toys still in existence. According to Cohen, Radio Rex is a charming piece of speech-recognition history but it has a terrible false-rejection rate. Like many of its flesh-and-bone counterparts Rex tends to remain stubbornly in its house despite the entreaties of its owner.
Julie and Jill
In 1987, Texas Instruments created a doll capable of responding to speech recognition utterances with spoken output. Described as the "first fully interactive" toy, Julie was a standard-size plastic doll that was indistinguishable from other dolls of the period except that it had a DSP chip that enabled it to respond to and generate speech. Julie could understand eight utterances: Julie, yes, no, OK, pretend, hungry, melody and be quiet. George Doddington, who led the team that created Julie, readily admits that Julie is prone to mis-recognitions that cause her to make inappropriate verbal responses. Around the same time Playmates Toys released the Jill doll. It was approximately 3 feet tall and had a number of mechanical features, including speech recognition and a removable three-track cassette tape. Jill would ask questions, such as "Would you like to play puppets, sing a song, or do a math puzzle?" Her word spotting grammar would then look for "puppets," "song" and "puzzle" in the child’s response ito select the appropriate track of the tape to play as a response. Jill’s speech-recognition abilities relied on the same chip that was embedded in Julie.
Toys in the New Millennium
The descendents of Radio Rex, Julie, and Jill include dolls, games, and action figures whose verbal skills have grown steadily in vocabulary, size and accuracy. Hasbro’s Aloha Stitch is a soft, blue and pink doll that can respond to 12 commands plus four other kinds of input that are part of its internal dialogues. For example, when it is asked to tell a joke, Stitch initiates its knock-knock-joke sequence. When the child presses the toy’s left hand, Stitch tells the child the list of commands it can understand. Like the movie figure it emulates, the doll can be happy or sad; rude or nice.
R2D2 is a technology toy for older children (including adults). Like its Star Wars namesake, R2D2 is a multifunctional droid. In addition to a speech recognition lexicon of more than 30 commands, R2D2 has a heat sensor that tells it when it is approaching a person who might want to give it commands, an RF sensor that can detect objects in its path, a motion sensor and battery-run wheels for navigation. It selects a heat source in its vicinity to interact with and uses a complex grammar that enables it to exhibit anger, annoyance and joy based on a previous event, such as hearing the name "Darth Vader."
R2D2 and Stitch are toys of the 21st century. They could be described as direct descendents of Radio Rex, Julie, and Jill but only in the same way that jumbo jets are descended from the Wright Brothers’ aircraft.
Dr. Judioth Markowitz is the associate editor of
Speech Technology Magazine and is a leading independent analyst in the speech technology and voice biometric fields. She can be reached at (773) 769-9243 or firstname.lastname@example.org.