Speech Connect for Cisco Unity

[Shawn Ford]

• Introduction
• Features and Benefits

Introduction

Cisco® Unified Communications Solutions unify voice, video, data, and mobile applications on fixed and mobile networks, delivering a media-rich collaboration experience across business, government agency, and institutional workspaces. These applications use the network as the platform to enhance comparative advantage by accelerating decision time and reducing transaction time. The security, resilience, and scalability of the network enable users in any workspace to easily connect anywhere, anytime, and anyplace, using any media, device, or operating system. Cisco Unified Communications is part of a comprehensive solution that includes network infrastructure, security, wireless, management applications, lifecycle services, flexible deployment and outsourced management options, and third-party applications.

The Cisco Unified Communications messaging portfolio provides three options:

• Cisco Unity®: Scales to meet the needs of large enterprise organizations and delivers voice, integrated messaging, and unified messaging options that transparently integrate with Microsoft Exchange, Lotus Domino, and Novell GroupWise.

• Cisco Unity Connection: Combines integrated messaging, voice recognition, and call transfer rules into an easy-to-manage system for medium-sized businesses with up to 7500 users. For organizations with up to 500 users, Cisco Unity Connection is available as a single-server solution with Cisco Unified Communications Manager Business Edition.

• Cisco Unity Express: Provides cost-effective voice and integrated messaging, automated-attendant, and interactive voice response (IVR) capabilities in select Cisco integrated services routers for small and medium-sized businesses and enterprise branch offices with up to 250 users.

Cisco Unity with Speech Connect allows people to quickly connect with their colleagues using only their voice. Speech Connect for Cisco Unity is a Speech-Enabled Automated Attendant (SEAA) for the enterprise. Internal or external callers speak the name of an employee into the phone and are connected to that employee. The Speech Connect solution delivers superior voice-recognition performance and uses enhanced software to store user preferences and fine-tune pronunciations, improving its accuracy over time. Users can collaborate more quickly and easily by connecting with each other without having to know multiple phone numbers. With the proliferation of mobile devices, flexible workspaces, and employee turnover, removing the need to know numbers greatly simplifies communications. If there are multiple employees with the same name or Speech Connect does not have a perfect match for the name spoken by the caller, Speech Connect will present a number of name choices for the caller and include additional information, such as the employee's location or department. This process is called disambiguation. Speech Connect will also play the person's recorded name in their own voice, making it easier for the caller to choose among the multiple names selected through the disambiguation process. Furthermore, Speech Connect performance improves over time through a self-learning mechanism that is included with the optional Names Tuning Service described in further detail below.

Speech Connect can be configured to respond to a number of different voice commands, not just employee names. For example, if you deploy product names in the Speech Connect grammar logic, customers can speak the name of a product and reach a person responsible for that product. Employee names, departments, conference rooms, product names, and any other entity that steers a caller to a phone number can enhance the value of Speech Connect to your employees and customers. While Speech Connect allows for faster, easier connections over the phone, Cisco Unity improves productivity by allowing employees to access messages anytime and anywhere, thereby enabling them to respond faster to messages.

Features and Benefits

Table 1. Features and Benefits of Speech Connect

Feature	Benefit
Support for up to 200,000 names	Supports the scale required by large enterprises.
Advanced disambiguation	Speech Connect stores caller preferences over time, expediting transfers as it builds knowledge about a caller's community-of-interest when given choices among similar names.
Names Tuning service	This service monitors results in an automated manner, inserting improvements and updated pronunciations within 72 hours of a failed utterance.
Speed dial	Speech Connect can be configured as a speed-dial soft key on Cisco Unified IP phones.
Name response collection	Allows the caller to hear the contact's name in the contact's own recorded voice to help determine proper pronunciation.

Note: Speech Connect for Cisco Unity is only available in installations that include a Cisco Unity or Cisco Unity Connection system.

Services: Speech Connect Names Tuning

As employees join and leave your company, there are frequent changes such as new names and pronunciations in your directory. To protect the performance and your investment in Speech Connect, Cisco provides an optional service called the Speech Connect Names Tuning service, which automatically corrects in the background, without you having to report issues to a technical support team. The Names Tuning service also includes professional linguists to correct problems relating to nuances in speech recognition - problems that cannot be solved automatically by any speech engine. This service not only fine-tunes the variety of employee name pronunciations within your corporate directory, it also provides support, monitoring, and reporting. The Names Tuning Service continuously improves the performance of Speech Connect to deliver superior, intelligent speech recognition capabilities.

A customer who subscribes to Speech Connect Names Tuning service receives:

• Enhanced linguistic support for 30 days post-installation

• Ongoing, automated monitoring and corrections (corrections generally appear within 72 hours of receipt)

• Cisco Help Desk support for recognition issues - the Cisco Technical Assistance Center (TAC) will prioritize any cases relating to linguistic analysis and correction and put you in contact immediately with a Speech Connect expert

• Monthly reporting of Names Tuning corrections and updates

• Alerts notifying your Speech Connect administrator of corrective actions to be taken

The Speech Connect Names Tuning service provides immediate benefits to your global enterprise. While it may be easy for employees to pronounce the name of their co-workers in their own country, proper pronunciation can be a real challenge when dealing with unfamiliar names from other parts of the world. The Names Tuning service monitors the performance of speech servers over the unified messaging network, correcting for pronunciation variances and errors within hours. This service helps ensure that Speech Connect is a valuable collaboration tool across borders and time zones no matter who is calling and no matter the name of the person on the other end of the connection.

Table 2. Speech Connect Names Tuning Service

Issue	Definition	Action
Out-of-grammar (entire name)	Speech Connect returns with no choices or presents incorrect choices.  Cause: the name spoken is not in the corporate directory, often because of timing issues with the accurate import of new employee records.	A Speech Connect Names Tuning service linguist identifies the missing employee record and reports absence to Speech Connect administrator.
Out-of-grammar (First name or last name)	There is an employee record for the desired contact, but results are poor because the first or last name has alternative pronunciations.	A Names Tuning service linguist adds nickname, preferred name, synonym, alternative pronunciation, or improved phonetic definition.
Low confidence score	Callers are being presented with many incorrect choices; there are too many choices to disambiguate. "Did you mean John Smith?"	A Names Tuning service linguist adds alternative pronunciations or improved phonetic definitions. If the poor results are due to audible voice quality issues, the issue is reported to the Speech Connect administrator.
Caller cancels or does not confirm	Caller is offered choices but says "cancel," hangs up, or transfers to the operator.	A Names Tuning service linguist adds nickname, preferred name, synonym, alternative pronunciation, or improved phonetic definition.

Product Architecture

This first release of Speech Connect runs on a separate Cisco speech server (Cisco MCS 7825, 7835, or 7845 Media Convergence Server) as an independent application. If you have deployed a speech server for Cisco Unity 5.x to take advantage of its speech-recognition features, another speech server will be required to support Speech Connect. Future releases of Speech Connect and Cisco Unity speech-recognition capabilities will merge onto a single speech server.

Table 3. Product Specifications

Product compatibility	Speech Connect is compatible with Cisco Unified Communications Manager versions 4.x, 5.x, and 6.x, Cisco Unity versions 4.x and 5.x, and Cisco Unity Connection version 2.x.
Software compatibility	There are no specific software requirements for Cisco Unified Communications Manager and Cisco Unity solutions because Speech Connect, in this initial release, operates independently. Company directory records can be imported into Speech Connect from any directory source, including Active Directory and Lightweight Directory Access Protocol (LDAP).
Protocols	Speech Connect supports SIP integrations.
Components	Speech Connect is an application that installs on a Cisco Unity Messaging Product Server for server (Cisco MCS 7825, 7835, or 7845 Media Convergence Server) running Redhat Linux and the Nuance 9.0 speech engine.
Ports	The speech server (Cisco MCS) is configured with the number of telephony ports required to support the desired call volumes based on the Cisco MCS server platform. See Table 4.
Integrations	Speech Connect will integrate to the local Cisco Unified Communications Manager or PBX over a SIP connection or through traditional TDM connections (using Cisco PIMG or TIMG).
Options	The Names Tuning service is the only optional SKU related to Speech Connect.
Performance	See Table 4 for name directory size and port considerations.
Programming interfaces	Speech Connect does not provide a programming interface, such as VoiceXML, for the building of voice applications on the same server.
Language support	Speech Connect supports U.S. English only.

Table 4. Supporting Hardware

Call Volume per Month	Company Directory Size	Recommended Server	Notes
1 - 20,000	1 - 5,000	Cisco MCS 7825	Will not migrate to speech server for future Cisco Unity releases
1 - 60,000	1 - 50,000	Cisco MCS 7835	Can be field upgraded to speech server for Cisco Unity releases
1 - 200,000*	1 - 200,000	Cisco MCS 7845	Can be field upgraded to speech server for Cisco Unity releases
Assumptions
Busy Hour (15%)	If you anticipate greater than 20% of daily calls to fall within one busy hour, increase configuration to next server level.
Steady Call Levels	If you have bursts of activity and calls are not evenly distributed over the month, increase configuration to next server level.
Blocking (P.01)	These recommendations assume potential blocking of 1% of calls during the busy hour. If you require a lower level of blocking, increase configuration to next server level.

* For higher call volumes, order multiple speech servers.

Cisco Unified Communications Services

Cisco and its certified partners can help you deploy a secure, resilient Cisco Unified Communications solution, meeting aggressive deployment schedules and accelerating business advantage. Cisco's portfolio of services is based on proven methodologies for unifying voice, video, data, and mobile applications on fixed and mobile networks. Cisco's unique lifecycle approach to services defines the requisite activities at each phase of the solution lifecycle. Customized planning and design services focus on creating a solution that meets your business needs. Award-winning technical support increases operational efficiency. Remote management services simplify day-to-day operations, and optimization services enhance solution performance as your business needs change.

Conclusion

Speech Connect enables a "numberless enterprise" where names are all that an employee, customer, or business partner needs to bridge the gap created by the proliferation of communications devices. Cisco provides the software and services to create this numberless enterprise, in addition to delivering speech-accuracy enhancements so that connections between logical people and teams become more accurate and instantaneous over time. In addition, Cisco Unity amplifies your productivity with an enterprise-ready voice and unified messaging platform as part of a unified communications solution. This system improves productivity by allowing employees to access messages anytime and anywhere, thereby allowing them to respond faster to messages for increased customer intimacy, service, and organizational agility. Cisco Unity with Speech Connect is an essential component of a unified communications strategy that enables collaboration across complex enterprises and across the human network.

For More Information

For more information about Cisco Unity with Speech Connect, visit http://www.cisco.com/en/US/products/sw/voicesw/ps2237/index.html or contact your Cisco account representative.