Multimodal Applications The Future is Now! Jean-Jacques Devaux Vice President Marketing & Business Development Telisma Frederic Dickey Director of Product Marketing Speech Products NMS Communications
NMS Innovators Webinar Advancing Communications with Voice Technology! Twelve months of telecom industry and technology information! Tailored for today s innovative solution developers! Presented by NMS and their partner application industry leaders! Complemented by hosted, on-line technology forums immediately following the webinar www.nmscommunications.com
Agenda! NMS " Multimodal! Definitions and background! Technology backdrop! Challenges! Telisma " Environment " Architecture " Solution! Q&A! NMS Developer s Forum open chat room www.nmscommunications.com
About NMS Communications! Leading provider of systems, system building blocks, and services for next-generation communications solutions " With a focus on enhanced services, voice-driven applications, and packet infrastructure markets! With a history of creating value for communications solution innovators equipment suppliers, application developers and service providers based on " NMS-built technologies and services " Strategic relationships with technology partners " Industry-leading supply chain and integration partnerships www.nmscommunications.com
NMS at a Glance Product Development Sales & SE support Channel Partners Headquarters! 20+ years in telecom! Products deployed in 90 countries! 24/7 worldwide technical support www.nmscommunications.com
Multimodality: A Definition! Multimodal applications combine speech, touch, data, graphics, and video into a single interface! Combine input (touch screen, keypad and voice commands) and output (images, sounds, text and speech) modes in a single device and user interface to allow users to access wireless content in the manner they prefer! Sequential vs. simultaneous www.nmscommunications.com
What?? Sample Multimodal Applications! Travel information " Make request via voice " Receive response in text! Directions " Make request via voice " Receive initial response in text " Get updates while traveling via voice or SMS or rich graphics! One-to-many messaging " Record message via voice or text " Deliver message via voice, SMS, WAP, or email www.nmscommunications.com
Business Drivers! Explosion of wireless and mobility " Up to now mostly for talking on the phone " Growth leveling-off! Key measures such as ARPU, subscriber churn! Wireless Internet " Handheld devices more and more powerful " 2.5G / 3G / Wi-Fi! Compelling new applications and services! Mobile terminals have small screens and keyboards " Difficult to use for input or output of complex information! Humans are multimodal! www.nmscommunications.com
Technology Backdrop! Multimodal means combining two different worlds " Voice network and the web! All-IP networks " VoIP transport and associated call (or session) control standards SIP, H.323, MGCP " Eventually One billion non-ip mobile handsets will take years to replace! All-IP networks not enough " Application development models must combine too! Voice apps Proprietary hardware, proprietary software, etc.! Web apps Completely open Voice application development must migrate to web model www.nmscommunications.com
Voice extensible Markup Language (VoiceXML)! VoiceXML 2.0 XML-based markup language for creating distributed voice applications! Used to create web pages with which a user interacts by: " Listening to spoken prompts text-to-speech or pre-recorded " Navigating via recognition of spoken or DTMF inputs! VoiceXML browser voice browser (continued #) www.nmscommunications.com
Voice extensible Markup Language (VoiceXML)! Benefits " Standard development environment " Decouples application from hardware platform " Applications portable across platforms " Taps into a very large development pool for Shorter development cycle! Restricted to voice (or DTMF) modality www.nmscommunications.com
Multimodal Markup Languages Two proposals at the W3C! XHTML + VoiceXML (X+V) " Combination of two well-known markup languages " Used to add voice to visuals for web developers " Linked together using XML Events to trigger elements of voice interaction! Speech Application Language Tags (SALT) " SALT tags extend existing mark-up languages such as HTML, XHTML, and XML www.nmscommunications.com
Where? Development tools Web Server (Web pages Server Side Code) Databases Voice Browser (voicexml or SALT) Telephony Server Natural Access API AG or CG HTTP Local or Remote Speech Engines Multimodal Browser (X+V, SALT) Multimodal Clients Speech Recognition and Text-to-Speech Sources: NMS Communications, W3C, IBM, Microsoft www.nmscommunications.com
Other Components! Distributed speech recognition (DSR)! Location! Presence and availability management www.nmscommunications.com
Market Challenges! Applications viewed separately " Voice services, text messaging, Internet access, graphics, streaming video, video conferencing! Lack of integration " Separate platforms for each service! Domain-specific developer tools! Walled garden mentality www.nmscommunications.com
Where NMS Can Help! Leader in open platforms application development! ASR, TTS and speaker verification are available on every NMS Convergence Generation and Alliance Generation platforms! Natural Access API " Speech is integrated with fax, conferencing " ISDN, SS7, and IP connectivity (continued #) www.nmscommunications.com
Where NMS Can Help! Tight integration with speech vendors " On-board advanced echo cancellation for enhanced speech recognition, DTMF, and tone detection " Barge-in detection during playback " Full-duplex capability for simultaneous play and record functions " DSP-based, voice activity detection (VAD) algorithm, significantly improving the overall performance of speech-enabled applications! Industry standards VoiceXML and SALT forums www.nmscommunications.com
Looking Forward! Big opportunities in wireless " Subscriber growth still " Emerging technologies " Infrastructure investment " Demand for versatile hand-held devices! Mobile Internet " Multimodal applications " Driving forces for the next generation of mobile communication systems www.nmscommunications.com
Telisma Highlights! Founded " August 2000 " Spin-off from France R&D! Headquarters " Paris! Product development " Lannion " Rennes! Sales offices " Munich " London! Headcount " 60 employees Lannion London Paris Rennes Munich www.telisma.com
Our Activity & Position! Editor of speech recognition software and development tools to enable mobile and largescale voice driven services! Provider of professional services to ensure business success of services deployments with our partners! Our customers telecom operators, voice service providers, large corporate accounts www.telisma.com
Carrier-Grade Products www.telisma.com
A Comprehensive ASR Environment Very Large Vocabularies Densifier Philsoft DGB Voice Distributed Framework MRCP server MRCP client 3rd party voice platform TTS connector! A comprehensive ASR environment offering highprecision recognition, optimized for large rollouts www.telisma.com
Densifer How it Works! Firmware loaded on telephony board offloads CPU, reducing data flow between telephony board and ASR client " Data stream reduction from 64 kbps to 5 kbps " Data flow reduction by an additional 20% in average (speech/noise detection) Audio signal Telephony board Speech/ Features noise extraction 64 kbps 5 kbps detection 5 kbps ASR client Dispatch MFCC decoding Densifying process www.telisma.com
Architectures www.telisma.com
Multimodality Types! Sequential GPRS terminals voice visual Operato Operato rr voice visual! Simultaneous GPRS class A or DSR on PDA voice a visual voiceavisual Operato Operato r r voiceavisual voiceavisual www.telisma.com
Scenario #1 Multimodal Access to Internet Services! Server-based only Visual Browser Voice/VoIP Voice Browser Internet Internet Internet Data Protocol Multimodal Synchronization
Scenario #2 Multimodal Access To Internet Services! Distributed Speech Recognition Voice Extraction DSR Protocol Voice Browser Internet Internet Visual Browser Internet Data Protocol Multimodal Synchronization DSR: Aurora www.telisma.com
Scenario #3 Multimodal Access to Internet Services! Terminal-based Voice Browser Visual Browser Internet Data Protocol Internet Internet Multimodal Synchronization www.telisma.com
Telisma DSR/Multimodal Solution www.telisma.com
DSR An Architectural Concept Features Featuresextraction, extraction, speech/noise speech/noise detection detection Acoustic Acoustic decoding, decoding, Speech Speechrecognition recognition Features Featuresextraction, extraction, speech/noise speech/noise detection detection Data channel (TCPIP, RTP, MRCP) ~5 kbps Acoustic Acoustic decoding, decoding, Speech Speechrecognition recognition Embedded DSR Voice channel (ISDN, PSTN, GPRS, GSM) 64 kbps, 8-15 kbps (VoIP) Features Featuresextraction, extraction, speech/noise speech/noise detection detection Acoustic Acoustic decoding, decoding, Speech Speechrecognition recognition Server
DSR Major Benefits! Improved speech recognition quality over wireless channels " DSR minimizes impact of speech codec & channel errors that reduce the performance from recognizers accessed over digital mobile speech channels! Multimodal applications enabler " Voice + data over a single wireless data transport rather than over separate channels, therefore, no need for a «class A» terminal allowing simultaneous use of voice + data signal! Guaranteed recognition performance level over every network " Same front-end, therefore no channel distortion coming from the speech codec and its behavior in transmission errors. and allows server-based pricing www.telisma.com
Telisma s DSR Strategy Prove DSR benefits! Application usability! Technical benefits sequential & simultaneous multimodality over GPRS, with proprietary DSR, enhanced VoiceXML offering, and Kirusa partnership Industrialize! Complete industrial Telisma + Kirusa + Sandcherry platform! ASR engine adaptations to support Aurora WI 008 normalized ASR client www.telisma.com
Unique Multimodal Solution Telisma s Distributed Speech Recognition! Truly carrier-grade for compelling applications! Broadest range of devices and network supported! Natural interface with simultaneous multimodality Show me the weather in Paris tomorrow. www.telisma.com
Architecture Multimodal Application Provider (1) Multimodal Application Provider (2) X+V SALT M3L Kirusa Multimodal Platform (KMMP) Telisma s Speech Solution GGSN Interface Markup Content Handler (X+V, SALT, M3L) Telisma voice distributed framework WAP Interface SS7 Interface SMS-C Interface Billing Interface OA&M Interface Simultaneous Module Client Handler User Database Sequential Module Multimodal Engine (State Manager) Device Database SMS Module Telisma Voice Browser Interface VoiceXML VoiceXML French ASR German ASR English ASR Telisma s Speech Solution Client French TTS German TTS English TTS 3 rd -Party Telephony Platform www.telisma.com
Benefits Philsoft DSR for Multimodal Solutions! Quality and reliability of the recognition trough the networks " Grammars accuracy! Vocabulary size: > 1,000,000 words! Natural language understanding, grammars complexity! Dynamic grammars management! Evolutivity and flexibility " Speech engine " Applications " Grammars tuning and updating! Terminal independence " Hardware and OS specificity www.telisma.com
France Trial! Simultaneous multimodality over GPRS in DSR " Form filling application " Kirusa partnership! Telisma contribution " Definition and experimentation of services " Study of the technical solution " Customer tests, trial support Jean-Jacques Jean-Jacques Damlamian, Damlamian, Group Group Executive Executive Vice Vice President President at at France France : : "We "We are are convinced convinced that that multimodality multimodality is is key key to to increase increase the the adoption adoption and and revenue revenue from from services services offered offered by by France France,, through through Orange, Orange, Wanadoo Wanadooand and Equant. Equant. Kirusa Kirusaand and Telisma Telismahave well well positioned positioned us us for for industry industry leadership leadership in in the the multimodal multimodal space." space."
In Summary Telisma is DSR-Ready! Product " Densifier, a DSR example already deployed " Many multimodal trials with telco operators " Exclusive multimodal event in VoiceXML offering for large rollouts or prototyping! Services " Significant experience gained through trials (user acceptance, project management, )! Partners " Kirusa, multimodal solution provider " Sandcherry for an industrial platform www.telisma.com
Please take a moment now to complete our short survey, while we start the Q & A
For more information! Contact " NMS Frederic Dickey frederic_dickey@nmss.com " Telisma Jean_Jacques Devaux jjdevaux@telisma.com! Register today at http://www.nmscommunications.com for the next two web seminars: " June 4 The Evolution of Messaging from IMM, MMS and Cell Broadcast to Video Presented by Imran Qidwai, Director of Messaging at NMS Communications " July 9 Enhanced Services in the Network and in the Enterprise
Solution Developer Forum! Join NMS and Telisma for live text chat on the Opportunities and Obstacles to Creating Multimodal Applications Tell us your story!! http://www.nmscommunications.com/chatpage " Log in instructions! Enter your user name (first initial, last name e.g., JDOE)! Password: nms! Check the scheduled discussion threads for more application/industry topics sponsored by NMS