SPEECH PROFILES: Voxware Brings Speech Recognition Into the Warehouse "Battlefield"

The recent acquisition of Verbex Voice Systems by Voxware Inc. brought Voxware’s president and CEO, Bathsheba J. Malsheen, Ph.D., back into the speech recognition field in a very big way. In February, Voxware, based in Princeton, NJ, which had previously specialized in speech and audio compression technologies, announced it had entered into an agreement to purchase the assets of Verbex Voice Systems, a private company based in Edison, NJ. Verbex is a leading provider of noise-robust speech recognition systems. Verbex’s core product is a Mobile Voice Pack (MVP), a belt-worn product that enables industrial workers to interact with warehouse and manufacturing tracking systems using voice commands. The MVP is specifically designed for workers who need to provide and receive information while their hands and eyes are busy, and when mobility is important. In explaining the acquisition, Dr. Malsheen said "we believe that there are opportunities for productivity-enhancing products in the industrial/warehousing markets and a real need for speech-based products exists, and their deployment can result in substantial savings to customers. We also concluded that the Verbex products perform remarkably well in the loud, demanding industrial environment." Previously, Voxware provided a number of technologies to software developers, hardware manufacturers, and service providers on an OEM basis. The company built voice and audio codecs, Intranet/Internet conferencing toolkits, and other PC-based solutions. But the purchase of Verbex and the company’s emphasis on speech recognition based solutions for industry makes Voxware very much an "industrial speech products" company. That brings the company in line with Dr. Malsheen’s experience. She joined Voxware in October 1996 with more than 14 years experience in the speech technology business. Prior to joining Voxware, she had served as the general manager of the technology business unit of Centigram Communications Corporation, where she built Centigram’s text-to-speech products. She holds a Ph.D. and MA from Brown University and a BA from Hoftstra. Shortly after the Verbex purchase, Speech Technology had an opportunity to interview Dr. Malsheen about Voxware and her vision of where her company and speech recognition are heading in the near future. Q. The purchase of Verbex and your subsequent effort to make Voxware a player in the speech recognition for industrial applications arena represents a repositioning for your company. Why are you doing it?
A. We see a very big opportunity for revenue growth. We see our market as the warehousing and industrial markets, not just speech recognition. Our focus is on warehouse picking, receiving, cycle counting, sorting, inspections, and inventory. This is a very big market, and it is growing. I think sometimes if an application does not seem sexy or cool, it can get overlooked. Warehouse management software (WMS) vendors are focusing on growth rates of 30% to 40%. There are a lot of special market dynamics in the WMS market that make this the right time for voice based solutions. After all, speech recognition was originally intended for eyes-busy, hands-busy situations, but the market dynamics were not really in place. Today, the situation is very different. People are used to the idea that you can carry around a computer. With factories becoming more automated, the efficiency with which goods move through the supply chain is more important than ever. Many companies are looking to reduce costs in the warehouse. Electronic commerce means more people are shopping by catalogue. That means more goods need to be stored in the warehouses and centralized distribution points. Plus, the costs of DSPs are continually dropping, and there is an acceptance of wireless infrastructures. Q. How does this market differ from the PC dictation market or speech recognition over the telephone? Does the industrial and warehouse market present specific technological and marketing challenges?
A. We are looking at a market here that poses harsh and difficult conditions for the speech recognizer. This has to be a very robust application and Verbex is known for the robustness of their applications. This is a definite requirement for voice applications in the warehouse. Noises from conveyor belts are often at a battlefield level. And customers need to be able to use the application for hours on end with very high accuracy. Besides needing to be a very robust application, any successful solution in this market has to function at a very high rate of accuracy. The people buying these solutions typically are vice presidents of logistics. Their job is to move goods, and any drop is noticed. A whole factory can be judged by how many packages get moved out of the distribution center. Because of that high accuracy requirement, we have a speaker-dependent solution, speaker adaptive technology, which can be adapted to anyone’s language. These are not people interested in trying a new technology. We don’t sell the newness of the technology. We make a return on investment case.
SpeechTek Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues