Solution Briefs

3 www.radisys.com RADISYS SOLUTION BRIEF | Integrated Speech Recognition Solution Powers Next-Gen Service Providers Critical Challenges in Providing "In-Call" Voice Enabled Services Costly and Complex Solutions Limits Service Innovation and Mass Market Outreach—The traditional approach to ASR deployment models, requiring external ASR servers, using high performance speech process engines, to process the entire gamut of speech interactions - from simple key-word or small vocabulary recognition to natural language and long form transcription. The cost made many applications prohibitive, limiting innovation in creating speech-enabled applications and reach the mass market. For instance, though the speech recognition technology with Natural Language Processing (NLP) capabilities are gaining high penetration in voice-driven interactive voice response (IVR) applications; due to associated hardware and software costs the use of such applications are many times limited to call centers (vs. everyday B2C and B2B applications) and accessible to only larger enterprises. Performance and Latency Issues Impacts User Experience—Sending and processing of all media files to an external recognizer results in unnecessary latency - particularly if the speech data is transmitted over the Internet to public cloud servers. Although machines are getting smarter and can learn better, the implications for the communications segment are much more sensitive. Inaccurate keyword recognition, processing, learning, analyzing and calling to action can initiate wrong processes that may end up in either a lousy user experience or even costing a substantial amount of time and money in business-critical applications. In the view of all the above challenges, CSPs and System Integrators need a solution partner that can provide integrated speech recognition service over their networks, coupled with the right media server solution to process and transform the speech to ensure an optimal user experience and improve recognition accuracy. User-A User-B User Call User Call MRCP Ex: IVR Dialog Application Server Always Required for Any Speech Processing (Basic and Advanced) Speech Processing Solution (Ex: Nuance) External ASR Server External Advanced Speech Analytics Media Server / MRF / Gateways T H E COS T O F T R A DITI O N A L A S R D E PLOY M E N T M O D E L S H A S P RO H I BI T E D S E RV ICE PROV I D E R S TO I N N OVAT E A N D R E ACH M A SS M A R K E T

• Cover