Last year Daniel Hong, voice business analyst at Datamonitor, predicted that the global voice business value chain would grow to $1.939 million by 2009 at a CAGR of 22 percent. This year he is forecasting that the entire market for global speech solutions will grow from 1.2 billion for 2005 to 2.7 billion for 2009. Speech Technology Magazine presents the fourth annual Most Innovative Solutions awards in recognition of the enterprises and service providers who are deploying innovative speech applications driving the growth of the speech industry.
To choose the 2006 Most Innovative Solutions, readers of Speech Technology Magazine and STM NewsBlast nominated companies who deployed speech recognition technologies to interface with their customers. A panel comprised of STM's Editorial Advisory Board selected the top 10 deployments who demonstrated creativity and innovation in their implementation of speech. The following are listed in alphabetical order and will be recognized at SpeechTEK 2006 in New York City on August 7-10. Many of these companies will be presenting during the conference.
Canon U.S.A. is a provider of business and consumer imaging equipment and information systems. Canon's speech solution is comprised of TuVox's Perfect Router and Aspect's CSS platform. The TuVox Perfect Router identifies caller intent and what product they are calling for and routes calls for Canon technical support, repair, products and accessories to the appropriate agent.
Canon's phone customer service challenges were to streamline the large number of incoming inquiries about Canon's diverse product lines and maintain its brand. Canon also wanted the agent to already be informed of whom the customer was and why they were calling so callers are never asked to repeat information.
With the speech application, the percentage of calls manually routed by live agents dropped from 42 percent with IVR to 27 percent. Adoption of automated phone customer service by Canon customers is rising with nearly 75 percent of customers choosing to route their call using speech instead of the traditional touchtone offering. Canon deployed 48 ports with server redundancy. The call center receives approximately 6,000 to 7,000 calls per day.
Comcast is a provider of cable, entertainment and communications products and services with 21.4 million cable customers, 8.5 million high-speed Internet customers and 1.3 million voice customers in 35 states.
Comcast was presented with a challenge: how to balance live agent and voice self-service support to provide the best possible customer experience and control costs. West designed, developed and implemented Comcast's natural language application. West Interactive worked with Comcast to design and develop a natural language-based troubleshooting application for Internet access, registration and installation issues. The use of SLM-based natural language enables a customer to speak a phrase or sentence describing their service issue.
This implementation represents the first component of a longer-term speech strategy for Comcast. The company plans to continue to work with its communications solutions partner, West, to speech-enable key areas within its IVR application. The natural language application currently processes 30,000 calls per day.
Health Net provides health benefits to approximately 6.5 million individuals in 27 states and the District of Columbia. Health Net implemented Aspect Software's Customer Self-Service (CSS) IVR solution in 2004 and upgraded to an enhanced version of the product in May 2005.
Health Net receives approximately 230,000 calls a month from customers needing to check eligibility, benefits, claim status, order literature or ID cards and more, all tasks that take more than a few voice prompts and a touch-tone menu. Using the Aspect CSS IVR solution, Health Net has realized cost savings of between $2.3 million and $4.4 million annually and achieve a 90 percent customer satisfaction rating for its automated system.
Health Net receives more than 2 million calls a year from members and providers. Of those calls, nearly 25 percent have been automated using speech technology. Of the 500,000 calls per year related specifically to claim applications, more than 60 percent have been automated via speech.
Landstar System delivers transportation services to a range of customers worldwide. Landstar's carrier group delivers over-the-road transportation services and Landstar's global logistics group provides international and domestic, multimodal (over-the-road, air, ocean and rail) transportation, and expedited and contract logistics services.
Landstar wanted to enable truckers to connect to the truck/load matching system using a tool they could access from the road—the phone. To add voice notification to its truck/load matching service, Landstar used Microsoft SQL ServerT 2005 database software, Speech Server 2004, and the Visual Studio 2005 development system. The solution enables truckers to receive voice notification of suitable loads over their cell phones.
Landstar plans to continue its work with Microsoft Speech Server, including use of the upcoming Speech Server 2007 platform. The system is averaging approximately 24,000 calls per business day running on 72 ports. There are currently more than 3,100 subscribers to the outbound notification service.
Nicor Gas is a natural gas distributor serving the northern third of Illinois. Each year, the contact center receives approximately 4 million calls. Nicor selected Nuance's SpeechPak solutions and Nortel's professional services team, with overall project management lead by AT&T. The SpeechPak was leveraged by Nortel's team to build a VoiceXML-based solution. Nortel's Media Processing Server 500s, Speech and Web servers provided the environment to house the applications that initially interfaced to Nicor's Unisys legacy environment and then repointed to SPL WorldGroup's Customer Care and Billing system.
Nicor's speech recognition business case and ROI specifically included the self-service enhancement experience of clients that migrated from DTMF solutions to speech recognition. Nicor has experienced a speech recognition rate of more than 90 percent since implementation.
Nicor deployed a 144 port Nortel MPS 500 system. The IVR handles approximately 450,000 to 500,000 calls per month, reaching a new self-service rate of 30 percent of all calls received, processed and completed within the IVR.
ParkingCarma (Acme Innovation)
ParkingCarma is the business segment of Acme Innovation. Acme Innovation launched ParkingCarma's SmartParking technology to allow drivers to determine the availability of parking spaces from their cars, homes, or offices. The technology was developed using Microsoft Speech Server 2004, Microsoft .NET platform, and Microsoft SQL Server. The use of speech provides an interface for drivers to navigate the menu and resolve their inquiry.
The ParkingCarma - Parking Guidance Service improved transit rider-ship by up to 14 percent, improved land utilization, and reduced VMT by up to 2 percent system-wide. For the pilot, 1,500 customers used the system tracking 50 spots in a parking lot of 1,000. Using this service, ParkingCarma increased utilization from 10 percent to 75 percent over 12 months. Customers use the speech system for these spaces at least 100 times per day. For this project, ParkingCarma used 24 ports of a 48-port board of which peak load was approximately five concurrent ports.
City of Chicago Department of Water Management (Teng)
The Chicago Department of Water Management (CDWM) contracted Teng to design and improve its contact center and to manage the solution. The CDWM has approximately 550,000 customer accounts and receive an average of 300,000 calls per year. The system will operate at the Chicago Department of Water Management call center in Chicago. The Department of Water deployed Nortel Media Processing Server 500 with Nuance Speech Recognition (ScanSoft OSR 2.1).
The contact center has the capacity for 30 customer service representatives; currently, there are 15 full-time CSRs handling all customer queries. The self-service system is a Nortel Media Processing Server 500 with 72 T1 ports. The contact center received over 300,000 calls in 2005; of which, a little more than 228,000 were handled by the IVR system. The average wait time dropped from 15 minutes to just over a minute and the average talk time was 3:10. The VoiceXML with full self-service automation was deployed in December 2005
The Ministry of Culture and The Ministry of Science, Technology and Innovation of the Kingdom of Denmark
The Ministry of Culture and The Ministry of Science, Technology and Innovation, on behalf of the Danish Government, have initiated a project to develop a speech recognition application for the Danish market. The application comprises three solutions: (1) Speech recognition software enables two broadcasting networks in Denmark, Danish Radio and TV2, to provide live subtitles of TV shows. (2) Speech recognition software which will be integrated within the parliamentary dictation system and used by law enforcement to create reports. (3) Speech recognition software to be sold to consumers on the Danish market. The solution used Philips Speech Recognition Systems: SpeechMagic, ConText, SDK Prolog Development Center, and Sysmedia.
The main goals were to provide broader audience access to computers and to deliver on the requirements of people with disabilities. The solution has an accuracy rate of 96-98 percent (measured in characters) and an average speed of 2.2 seconds from the time the last word in a sentence is spoken until it appears on the TV screen.
Transcend Services is a provider of medical transcription services to health systems, hospitals, clinics and physician practices. Transcend selected AnyModal CDS by M*Modal to improve the company's operational efficiency by increasing the number of lines of transcription a medical transcriptionist can produce within an equivalent amount of time; reduce the turn around time for physician dictation by producing an edit-ready structured clinical document in the background while medical transcriptionists work on other documents; and improve the quality of medical transcription reports by allowing the transcriptionist to focus on document accuracy instead of typing and formatting.
In addition to using M*Modal's speech platform, Transcend integrates M*Modal's transcription editing component into the BeyondTXT user interface, providing a seamless interface for the transcriptionist. AnyModal CDS has become an integral part of Transcend's improvements in transcriptionist productivity and operational efficiency.
M*Modal processes several hundred hours of dictation per day for Transcend. The solution is used to prepare edit-ready clinical documents for more than 40 medical specialties.
Vodafone Spain forms part of the Vodafone Group, a mobile telecommunications services provider with a presence in 27 countries, and agreements with 32 others throughout five continents. The Vodafone Spain ASR IVR was developed and designed by Ydilo Advance Voice Solutions using their PeopleBase platform, which uses Nuance's speech recognition engine.
Vodafone Spain deployed 1,800 ports in two different sites (Madrid and Barcelona). Currently the IVR serves more than 250,000 daily calls, though it has peaked 330,000 during the Christmas campaign, managing 62 percent of those calls completely in the IVR. The solution served 51 million calls in 2005 and 21 million calls in early 2006. Vodafone Spain has approximately 11,500,000 clients using the IVR (88 percent of Vodafone's clients). Vodafone Spain also has an average of 13,000 surveys performed daily by the IVR. The system has also enabled more than 750,000 prepaid customers to join the company's Loyalty Points Program, more than 280,000 gift redemptions since 2004, and more than 1,100 pricing plan changes performed daily.
*The figure reflects global spending on speech platforms, enabling software, applications and services from 2004 to 2009. It does not reflect spending on hosted speech revenues outside of applications and consulting/systems integration services.