Japan's Researchers Working on Natural Sounding Robotic Speech

Researchers at Japan’s Waseda University are working on robots capable of more natural speech and interaction with people. The potential applications of such advancement include reduced bandwidth load for mobile phone communications, improved control over artificial vocal cords for people unable to speak and better tools for speech training and language learning.

The basic goal of researching speaking robots is to understand how the human brain controls speech actuators when people talk. Waseda University computer scientist Masaaki Honda said, “What we do not clearly know is … how the different circuits in the brain work together to produce speech sounds; and we won’t understand it exactly until we can reconstruct the brain circuitry and machinery of speech.”

Honda created the Waseda Talker, a robot that produces human-like speech sounds by pushing compressed air through an artificial vocal tract equipped with motors that move the lips, tongue and vocal cords. Teeth and a nasal cavity have also been outfitted into the device. The increasing flexibility of its palate, tongue and lips are the key elements in the Talker model’s development. The Talker could generate more natural-sounding speech with the addition of elements such as protruding lips and a new control mechanism for the vocal cords. The sound analysis software helps the robot to copy humans and mimic certain words. In the future, the machine might mimic words on its own, once researchers have developed a computer model for voicing phonemes.

Source: New Scientist