Speech in the Network
Business Outline
ANSR^TM

INTRODUCTION

Sprex is a leading Seattle-based speech recognition technology vendor focused on "Speech in the Network"^TM. Our core technology is on the cutting edge in this field; we are developing an internet-based infrastructure for anytime/anywhere speech-based interaction with arbitrary products and services via audio-enabled, network-connected devices. With cash on hand for 1 year of no-growth operations and 5 purchase orders for OEM licenses received, Sprex seeks Series A funding from first-tier VC to support growth.

BACKGROUND

Microsoft recently paid $90M to acquire Entropic, the technological leader in speech recognition technology (per frequent NIST/ARPA benchmark comparisons) -- and a supplier of Unix speech rec technology to Sun. To minimize lawsuit risks from Entropic's Unix OEM's, who had invested in Entropic's now-withdrawn technology, they licensed the software needed to build speech recognition based products and services to a small consultancy, Sprex, Inc., so as to support those OEM's, and to create & exploit other derivative works, including engine source code and other tools for Sprex-internal use. Sprex instantly became technologically equal to those vendors with an independent code base for state of the art, practical speech rec systems, notably IBM, Philips, L&H/Dragon, and Nuance. Sprex intends to fully exploit this opportunity and has shifted into startup mode, with the intent of becoming a major speech technology vendor, by establishing the speech platform of the future -- Speech in the Network -- and dominating its long-term commercial exploitation.

PRODUCT

Sprex has already implemented a partial, working preliminary version of its core product ANSR (the Automatic Networked Speech Recognizer). ANSR is a network-based speech recognition SDK with commercially licenseable components that enable an easily integrated, seamless, minimum-latency, anytime/anywhere means of control of and interaction with arbitrary network-connected devices and services using natural spoken language.

CUSTOMERS

Providers of mobile, wireless data and voice services, internet based services such as voice-enabled websites, product or service user interfaces, and home-networked devices will buy the ANSR infrastructure and Sprex customization services to enable automatic, voice-based end-user interactions. Nuance/SpeechWorks VARs will also buy ANSR as a competitive technology alternative for speech rec enhanced call center systems (with 65K centers worldwide employing 2.8M staff, Nuance leads with fewer than 100 applications deployed so far). We have the option to be a follower in Nuance's telephony-ASR market, but we intend to be the leader in the unclaimed networked-ASR market. For near-to-medium term income, Entropic speech rec customers (the majority of serious speech R&D labs in the world, but fewer than 1000) can, for the first time, deploy their Entropic-based applications commercially by sublicensing Sprex' commercial use rights in the form of ANSR licenses.

BUSINESS MODEL

Sprex may host public and private ANSR recognizers on a server farm. The private servers will provide guaranteed capacity for per-minute connect charge (pennies per minute); public servers will provide free capacity for development, demonstration, and testing, but generally will be too slow for satisfactory after-deployment operation. Tools will be available for free to developers at least sufficient to build minimal applications, working with the public servers. Standalone or additional tools may be sold for a price. ANSR engines can also be licensed for local installation at business-software prices or leased for local operation with per-minute charges. Pricing may be pegged to quality of service measures, so that systems with improved response time cost marginally more. We aim for a cashflow model primarily based on connect time, whether to the Sprex-hosted private servers or to customer-hosted servers, although one-time fee deals are possible.

TOM VEATCH, CEO.

'84 BA and '91-'93 Visiting Professor in Linguistics at Stanford University; U Penn PhD, specializing in conversational-style speech R&D; 17 yrs in speech R&D; 6 yrs sales & tech consulting in speech recognition. I will be a strong CTO when Sprex is a major speech rec vendor; meanwhile I wear many hats. As former sales rep for Entropic, I have unparalleled hiring contacts among speech engineers. Several PhD- and Master's-level speech and telecom engineers are working on the product, part-time, for stock, or available to bring in once funding is available.

NEEDS

We seek first-round funding to execute on our plan, expand Sprex' development efforts and team and to establish the network-based speech recognition software market. We estimate total HR costs for the next 30 months to be $16,133,700. Investor assistance is sought for high-level staffing, strategic partnerships, and interfacing with the financial community in later funding rounds and at IPO.

RETURNS

Sprex aims for long-term dominance of a major speech rec platform and market. The small number of independent speech rec providers, the unusually high technical barriers to entry, the wide utility of speech, and the focus on speech in the network for both competitive uniqueness and the leverage of a powerful driver of growth, all these are good reasons to think we will succeed in our aim. If successful, we will grow to IPO and beyond, and provide significant capital appreciation to our investors, with liquidity as IPO conditions ripen, probably after about 3-5 years.

Comments?
If anything can be changed to better address your wishes, we are eager to hear about it. Please share any reactions you may have.

   From:
Message:

Speech in the NetworkBusiness OutlineANSRTM