Speech Profiles: Speech Recognition Chips 101— It's Serious Business Now

Recent advances in the area of low cost speech recognition have moved the technology into everyday consumer products. With vast improvements in the quality and accuracy of low-cost speech recognition systems, the value of adding speech recognition to everyday consumer products is now being realized. As products become increasingly sophisticated and offer more functions, implementing speech recognition allows consumers to use products more intuitively while maximizing their functionality. Talking to our products and listening to what they have to say gives products a life of their own and significantly changes the way we can use them. One area in which speech recognition will have a deep impact is voice dialing. This allows consumers to dial a phone number simply by saying the name of the person they wish to call. Sensory Inc. first made its reputation as a company that made toys talk. While enabling teddy bears to speak may seem like a prosaic corporate goal, it requires bringing maximum functionality onto one low cost chip, and pursuit of that goal has made Sensory an industry leader in fields of neural network, speech synthesis, and integrated chip technology. As products become more sophisticated and offer more functions, speech recognition affords consumers the opportunity to use products intuitively. Soon, using such products will not involve a series of button pushes and beeps, but actual dialogue between the user and the product. Sensory recently announced Voice Dialer Chip, which lets designers integrate voice dialing into existing products. In January, Sensory chips were instrumental in the launch of the world’s first voice activated cordless telephone, the EXV98 VoiceDialer phone from Uniden. Sensory was founded in 1994 by Todd Mozer, president of Sensory, his father, Dr. Forrest Mozer, a physics professor at University of California at Berkeley, and his brother, Dr. Michael Mozer, a professor of neural networks at University of Colorado at Boulder. The Mozer family developed the first speech synthesis chip over 20 years ago. Today, Sensory is the industry’s largest supplier of dedicated speech recognition chips, with about 85% of the market. The company develops and commercializes technologies, speech among them, that enhance products with communication. The InteractiveSpeechTM line consists of single chip processors intended for low-cost consumer electronic applications. Speech Technology recently interviewed Todd Mozer about Sensory, the company’s plans for the future, and his vision of where the speech recognition industry is heading. Will the lower price points on the chips mean speech recognition will become more of a mainstream product with consumers?
Yes. I think the announcement of the first voice activated cordless telephone by Uniden America was the first step to bringing speech recognition technology to the consumer mainstream. The EXV98 is a very exciting product, and there is a major campaign behind it. Uniden is giving it a real push and they are the world’s largest manufacturer of cordless telephones. Sensory has put a lot of focus on the voice dialing telephone handset. Speech is a major leap forward for the telephone. It is a major transition stage, just like when the rotary dial replaced dialing the operator. What impact will speech have on how we use the telephone?
You’re going to see padless telephones in the future. The keypad will go away. Within five years voice dialing will be as prevalent as speed dialing is now. You won’t even need it to punch in your friends’ phone numbers. You will even program the pre-set numbers by voice. What sort of changes can we expect outside of the telephone area?
You’ll see more consumer electronics products controlled by voice. Customers will use their voice to control light switches, refrigerators, PDAs, (personal digital assistants) and many other products. Would you categorize Sensory as a semi conductor company or a software company?
We’re the only company offering speech recognition technology both as dedicated (proprietary) chips and (proprietary) software. But ultimately, our real game is in high volume markets. Consumer (electronics) markets. We are the only company offering speech recognition technology that offers both chips and stand alone software. Our focus is on speech recognition. Even though we have introduced biometric and other voice recorded products, we are a part of the speech market. Continuous dictation gets a lot of praise, but we have not put our focus there. We have focused on the consumer segment. As more and more consumer products are controlled by voice, I think Sensory will be in an excellent position.