![]() ![]() |
||||||||
|
|
|
|
|
|
|
|||
|
Product
News
Product News
Corporate News
|
|
|||||||
|
Dr. Ami Moyal, CEO of Natural Speech Communication (NSC), has a vision. One day, he believes, we will stop pressing buttons on machines and simply tell them what we want them to do Take the VCR, for example, a machine that even Moyal, an engineer, admits he does not always know how to use. "We should just be able to say that we want to record a program at a certain time for the machine to work," says Moyal. "Why should I push any buttons? Any communication between human to machine should be by voice. Machines should understand our language." It is this vision that has guided Moyal, an expert in speech recognition, throughout his 18-year career. Today, his company, NSC is taking the first steps in this direction with Automatic Speech Recognition (ASR) engines designed specifically to replace human agents or interactive voice response systems at telephony and call center industries. NSC, which is based in Rishon Lezion, has developed a range of solutions that enable application developers and system integrators to provide robust voice-driven services to telecom operators and service providers. The NSC solution is based on PCI boards that allow very high density support - serving thousands of users simultaneously - at a low cost, and with low power consumption. The boards, which can be configured to support multiple applications and languages, perform all the recognition operations without the need for CPU resources, and eliminate the need for additional computing resources. The company began selling its products in the second quarter of last year, and is now seeing revenues increase quarter by quarter. By the end of next year, the company hopes to break even, and by the end of 2006, expects revenues to be over $10 million. NSC was founded in 1994 by two former employees of the DSP Group. Until 1997, the company was a garage operation with just four or five employees bootstrapping while they developed a speech recognition technology. In 1997, telecommunications company Telrad Networks invested $2.5m., and the company was finally in a position to dedicate itself to developing the technology. In 1999, a second round was held, bringing in the venture capital funds, Polar Communications, an investment company traded on the Tel Aviv Stock Exchange and Corex Israel Industries - a consortium of South African shareholders. AudioCodes, a specialist in the Voice over IP market, joined the round at the end of 2001, bringing the total amount of the round to $4.5m. The four shareholders have invested a further $2-3m. since then. In 2001, NSC ran into trouble. Performance was poor, and the company's burn rate was high. In May that year, the shareholders fired the founders, cut the burn rate, brought on Moyal - a former NSC employee - as CEO, and restructured the company. "Our employees call us NSC B, because we're like a completely new company," admits Moyal. SINCE THE restructuring, company development has moved on at a faster pace. At the end of 2002, Israeli mobile operator Cellcom began testing NSC's technology. A year later it launched a speech-based corporate address book for Cellcom's 4,500 employees using the technology. In April this year, the operator launched its first public speech CRM service for its 2.3 million subscribers, based on the NSC ASR engine. The service provides roaming information and pricing in English, Hebrew, Russian, and Arabic. Subscribers can get information about roaming services and rates just by saying the country name. "This new service is part of Cellcom's efforts to continuously improve customer service and to provide its subscribers with more convenient and user-friendly interfaces," says Yaron Horowitz, VP of customer service at Cellcom. The only problem with the telephony market is that the sales cycle is long and slow. Cellcom tested the system for over a year before making a purchase. The call center market has a much shorter sales cycle. Currently 10 NSC systems are being used in call centers across Israel. The company's main marketing goal is to be integrated into other technologies and platforms, and it is hunting for OEM and reseller agreements. In July 2003, the company signed a marketing agreement with IBM Voice Systems, to support the IBM Websphere voice response platform with a high-density speech recognition solution. The platform integrates information from multiple sources and delivers direct access to services and information 24 hours a day. NSC has also integrated its technology with five or six other Israeli companies, including ITS Telecom, which develops and markets peripheral telephony software, several in the US, and more in the UK. It also recently signed a distribution agreement with AudioCodes, which Moyal believes will help boost NSC's sales staff, which currently consists of just a few people. NSC is now training sales staff at AudioCodes to sell NSC products. In July 2002, the company signed a partnership agreement with Persay, a spin-off from Comverse that specializes in the complementary field of voice verification systems for remote services. NSC incorporated Persay's technology to its speech recognition board so the companies could offer users both speech recognition and text to speech recognition. The companies are now cooperating on both the marketing and technology levels. Persay's backers include Polar Communications In future, Moyal believes the vast majority of NSC's income will come from large resellers. ONE OF the advantages of NSC's technology is that it also has applications for the rapidly emerging homeland security market. When NSC developed its technology it did so in such a way that the system could recognize key words in a sentence, and act upon these words. "People are not always answering the question they were asked by the answering service so we were obligated to develop the capability of extracting this information from a sentence," explains Moyal. "When people use an answering service they want to speak freely. They don't just say the name of the person they want to contact, they say, "Please could I speak to John at home," for example. We have developed the capability for our system to adapt to normal human behavior and to pick out the key words in this sentence, John, and home." The upshot of this is that the technology can also be used to pick up any key words from telephone conversations, including words like "bomb," or "explosives." "Homeland security is a basic need today," says Moyal. "The authorities must to be able to monitor conversations to locate information, but this is currently a very expensive task. Our technology makes it possible at a much more reasonable price." The pressing need for such technology means that this area of business is moving very quickly for NSC. Moyal believes that NSC will sign some significant deals in this sector in the coming months. This technology also has a place in the business world, where companies want to monitor and keep track of calls coming into customer call centers. NSC's biggest competitors in this field are two Californian
companies, Nuance, and ScanSoft. Nuance, which employs over 300
workers, and saw sales of $55m. in 2003, sells mostly in North
America, but is also making itself felt in Europe and the Middle
East as well. The company hopes to reach profitability during the
second half of this year. Together, these two companies dominate the worldwide market with a combined market share of over 50%. THOUGH THESE companies are strong and NSC is just a start up, Moyal believes that NSC offers many unique features that will help transform speech recognition into a standard tool. For a start, unlike both ScanSoft and Nuance offer just software-based solutions. NSC also offers users high density, supporting a much higher number of simultaneous calls per box than competing speech solutions, according to Moyal. This lowers the solution cost, and makes for much simpler deployment. NSC also believes firmly in competitive pricing. "If we want a technology to be widespread, the product should be priced appropriately," says Moyal. The traditional pricing model is an expensive system whereby software is sold with a license per channel. NSC, however, sells its boards at a fixed price, and these boards can run on any number of channels the vendor wants. "We are leading another pricing method in the market," says Moyal. "The feedback from the market is very good." Moyal is also confident that NSC's technology is better than its rivals. "Cellcom tested our system and competing systems in very severe conditions, but they decided to work with us," he says. "All the technologies on the market today are good at some things, and not so good at others, but as the market moves from early adopters to acceptance, market performance is not the only issue. So is price." At present, NSC is not looking for additional investment. "Our four shareholders are backing us," Moyal says confidently. "We don't need external money." The company, which currently employs 22, may start looking for cash at the start of next year through a private offering. It has offices in New York and the UK, and wants to open more in the US and the Far East to increase its global presence. Even once sales have begun properly, Moyal says he does not expect the company to grow too large, ideally with just 40 or so staff members. "We are preparing ourselves to be an OEM company," he says. The company's goal now is to start large-scale deployment of services, and to start expanding globally. In the long-term, Moyal's aim is to take a double figure share of this rapidly emerging market. While Moyal realizes that the company's competitors are no doubt also exploring the homeland security market, and other applications in the future, he still believes that NSC has something unique to offer. "Until now, companies have not deployed speech recognition technology because it is too expensive," he says. "We aim to turn speech recognition into a commodity."
|
||||||||
| Iridiacom Group :: Telrad Connegy :: Sitemap :: Disclaimer :: Privacy | Copyright © 2005 Iridiacom Ltd | |||||||
|
|
|
|
|
|
|
|
||

