Dr. Ami Moyal, CEO of NSC

Natural Speech Communication Ltd. (NSC) is a developer of speech recognition products for telephony speech communication applications. NSC is a privately owned company, backed by companies and VCs including AudioCodes, Corex Israel Industries, Telrad Holdings and Polar Communication.

NSC is involved in the implementation of speech recognition based applications through business partners: IVR and VoiceXML platforms, telephony application developers, system integrators, and recording platforms vendors. The company is active in North America, Europe and Middle-East through local offices.

Speech Technology Magazine interviewed Dr. Ami Moyal, CEO of NSC to discuss current issues.

Q. Dr. Moyal, NSC recently launched the NSC SpeechBlade™ Server - why is this product unique?

A. I will start with a short description of the product than describe its unique benefits for the end users. The NSC SpeechBlade Server is a speech recognition server for telephony communication applications. The product includes multiple speech processing engines operating on a single dedicated platform.

It currently supports two speech recognition engines:

• The NSC Spotter - a robust word spotting engine with real-time capabilities, and
• The NSC Speecher - a speech recognition engine designed for telephony speech driven services.

The NSC engines operate on a dedicated platform, The NSC SpeechBlade, which is family of PCI blades that enables the entire speech processing operation to be done on the blade.

We are looking here at a non- traditional approach in the market. We offer a product that provides the speech recognition technology coupled with a compact, high density, platform for operating the speech recognition engine. This approach allows a "CPU free" product, which actually includes all the resources needed for the speech recognition operation. The approach also enables very simple system architecture, since adding speech recognition capabilities to a system needs only free PCI slots anywhere in the system.

The product also provides its users with flexibility to control resources where different engines, resource types, grammars and languages can run simultaneously over the same platform - the PCI blade.  This flexibility is coupled with the ability to have up to 120 independent speech recognition resources on a single PCI blade, serving multiple applications.

Q. Along with this announcement, NSC presented a new business approach to building and deploying speech recognition products. What is this new approach and how can it impact the "traditional" ways of building these products?

A. The new business approach NSC is presenting is a direct result of the NSC SpeechBlade Server capability of providing the end customers with "Speech Processing Power" that can be used by multiple speech processing engines. Our product offers the flexibility to easily adjust the use of these resources according to different business priorities, on daily base. In addition, NSC pricing model is per PCI blade and nor per licensed channel.

For example, let's consider the speech processing needs of a modern call center. It will need speech recognition for automated speech driven interaction with callers, word spotting for both speech analytics and quality improvements and speaker verification for authenticating caller's identity. In today's reality, the call center will need to purchase a separate product for each of these capabilities. In most cases, separate infrastructure and resources will be used in order to support each capability. With the approach offered by NSC, the call center can purchase a single product supporting simultaneous operation of multiple speech processing engines on its dedicated operation platform, managed through a single interface. The call center will be able to determine how and when to use its resources, as the business requires. For example, during the day, you may want to use speech recognition and speaker verification for customer support, while at night, using the same product, you may use key-word spotting for speech analytics.

Q. What impact will this new business approach have on speech technology's end user customers? What effect will it have, if any, on your current/past customers?

A. I think the main impact on the end customers will be the fact that the deployment  and use of speech processing technologies will become easier, need less resources and  cost much less. Customers will be able to maintain their independency and flexibility in operating their speech applications as well as present an economical business model and roadmap to adopt speech technology in their business. The approach of not constraining the end customers and giving them the freedom to configure their resources flexibly will result in higher customer use and satisfaction.

Q. Speaking of customers, Cinema City - a cinema complex with 23 cinema theatres in Israel - has recently deployed a solution using NSC's speech recognition engine. Please tell us about this deployment.

A. Using the new service, the caller is able to say the name of the movie he or she wants to see, choose the time and date of the show and the number of tickets required. The ordering system performs the activity without the caller being asked to browse through touch-tone menus during the call. The new service is a 'first of its kind' in Israel, and shortens the ordering time over the phone. It also makes the process more user friendly by providing the caller with an intuitive interface that has a reduced number of stages. This ticket-purchasing service can also be used when the customer is 'on the go', using the mobile, or even while driving a car ("hands free"). I think there's a real market here for ASR applications and I hope similar deployments will follow.

Q. Do you have more new deployments in the works that you could tell us about?

A. Recently, NSC's speech recognition product was deployed in a stock quotes service at Israel's leading bank. We have several active speech driven services with Israel's three leading cellular operators, which are implemented in their call centers as well as in their value added services. Outside of Israel, NSC sells through distributors, applications developers and system integrators.   NSC has also deployed its word spotting engine in some projects for the security market and call centers.

Q. Does NSC have any other projects in the works for 2006 that would interest customers and partners?

A. I believe that in 2006 NSC will continue its growth in sales and deployments. Part of the projects and deployments will be public but, in some cases, such as in the security market, the activity will not be published. I believe we will see more embedding of NSC's blades into target platforms, such as PBXs, for which CPU resources are not always available for speech recognition and therefore, NSC's approach has a clear advantage.

As for our R&D, NSC plans to continue to launch more speech processing engines that will operate as part of the product over the same dedicated platform. By that, NSC will broaden its offering and the customer's capabilities of operating a multi-engine speech processing environment that is economical, with minimum technical efforts.

Q. Is there anything that you would like to add?

A. First of all, I'd like to wish all our colleagues in the speech business a fruitful and successful 2006. The market is progressing every year and I strongly believe that this progress will continue until eventually we will look back and wonder how we managed without speech recognition.