Installation and Administration Guide for the Speech Connect for Cisco Unity Feature
Names Tuning Service Administration
Downloads: This chapterpdf (PDF - 102.0KB) The complete bookPDF (PDF - 3.01MB) | Feedback

Names Tuning Service Administration

Table Of Contents

Names Tuning Service Administration

Names Tuning Overview

Installing the Names Tuning License

Scheduling Names Tuning Tasks


Names Tuning Service Administration


This chapter describes the Names Tuning services process, and explains how to use the Speech Connect administrative interface to set up secure FTP communications and to schedule Names Tuning tasks.

See the following sections:

Names Tuning Overview

Installing the Names Tuning License

Scheduling Names Tuning Tasks

Names Tuning Overview

Revised February 2008

The Names Tuning service provides continuous improvement for your Speech Connect grammar and dictionary. It provides up-to-date pronunciations for all names, and resolves common nicknames (for example, Mike for Michael), reducing errors and ensuring the success of your system. The Names Tuning service produces periodic reports that show the results of the monitoring, analysis, and modifications to the grammar.

The Speech Connect Names Tuning sentry program logs each call to capture recognition results, the actual utterances themselves, and the actions taken by the caller after being presented with recognition results. It builds a transaction file for Names Tuning processing. Transcription logs are reviewed for accuracy and modified as necessary. Adjusting the phonetic representation of a name is a valuable part of the Names Tuning service.

The Names Tuning process updates your Speech Connect directory with new employee information and pronunciation definitions. Your system sends utterance files from the call logs to the Names Tuning service for transcription and analysis. The transcriptionists assess the overall recognition accuracy and trained linguists make any necessary phonetic adjustments based on the actual spoken utterances of the caller. The adjustments update your system master phonetic dictionary, the Linguistic Definitions Dictionary (LDD). In response to a failed utterance, the transcription may indicate the need for a nickname or a new employee or department record.

The Names Tuning service sends e-mail periodically to notify the administrator that the following reports are available:

Client Recognition Summary Report

Failed Utterance Report

For more details about these reports, see the "Names Tuning Report" section on page 9-6.

Installing the Names Tuning License

When you register for the Names Tuning service, you receive a Names Tuning License. The Names Tuning Licensing fields allows you to:

Enter your Names tuning license information.

Verify the outbound sFTP connection to the Names Tuning server.


Note For Names Tuning communication, you must provide secure network access from within your local network to the internet. All network connections for the Names Tuning service are initiated from behind your firewall. You do not need to allow inbound connections from the internet to Cisco for the Names Tuning service; however, depending on the general support arrangement, your technical support representative may require access from a remote internet connection for general maintenance and support.


To Register Your Speech Connect System for Names Tuning Communication


Step 1 In the Speech Connect interface, click Config and then click Global Settings. The Names Tuning Licensing section appears on the Global Settings page.

The Names Tuning Licensing fields are described in Table 8-1.

Step 2 In the Install License File field, click Browse to find and select the license file that was sent to you for your Names Tuning registration.

Step 3 Click Install License to install the Names Tuning license on your Speech Connect system.

Step 4 Click Test URL to verify the outbound sFTP connection to the Names Tuning server.


Table 8-1 Names Tuning Licensing 

Field
Description
Values

Information/Status

Shows the expiration date of your Names Tuning registration.

System displays the expiration date in MM/DD/YYYY format

Install License File

Allows you to browse your local drive or network to select the license file that you received for your Names Tuning registration.

License file

Names Tuning URL

The URL that you received from your Names Tuning registration package.

System displays Names Tuning URL. For example, ftp.domain.com.


Scheduling Names Tuning Tasks

Revised February 2008

After your employee and department records are imported, the Rebuild Grammar process enables the recognition engine to compile the records into a speech recognition grammar by building pronunciations from a locally stored image of the LDD. With the first running of the Names Tuning process, your names directory will be updated with any new definitions from the LDD.

As you add and update user records, the new records are queued for processing by the Names Tuning process. The updates may be in the form of imported transaction files, the entire employee directory, or changes that you made directly to the directory through the Speech Connect interface.

The Names Tuning process compares the update files and sends all changes to the Names Tuning service for processing. After the Names Tuning process has completed, your Speech Connect system retrieves the updated dictionary file and synonyms. These processes are entirely automated and complete without any human intervention. Upon completion of all the Scheduled Tasks, the directory update service sends an e-mail to the Speech Connect administrator that shows the status of the Names Tuning and Directory Update processes.

Scheduled Tasks set the frequency and start time for automated processing of the following tasks:

Send Transcription Logs—Sends the transcription logs and utterances to the Names Tuning service. The logs are text files that log all spoken names, and the utterances are recorded for each call.

Retrieve Dictionary—Retrieves the updated dictionary from the Names Tuning service. The dictionary contains a list of all words in your grammar, synonyms, and phonetic pronunciations.

Get Synonyms—Retrieves the master list of nicknames (for example, Jim for James), and loads the local database.

Send Word Lists—Sends the word list to the Names Tuning service. The word list contains every word in your grammar.

Data Load—Processes data files that were imported into the system by using the Speech Connect interface or sFTP. For more information about these processes, see the "Importing Data from Multiple Sources" section on page 4-1.

Active Directory Load—Performs the Active Directory Import tool automated process, which queries Active Directory for employee data updates and imports them into the Speech Connect database. For more information about this process, see the "Importing Active Directory Employee Data" section on page 4-4.

Rebuild Grammar—Updates the speakable directory with employee and department data. The process also creates a file for input to the TTS tool, which contains a list of the employees who do not have a spoken name or TTS recording. The TTS tool takes that list and generates TTS for those employees as well as departments and tiebreakers.

To Schedule Names Tuning and Directory Update Tasks


Step 1 In the Speech Connect interface, click Config and then click Scheduled Tasks.

Step 2 Enter values into the fields, as applicable.

The Scheduled Tasks fields are described in Table 8-2.

Step 3 Press Update to update your Names Tuning tasks schedule.


Table 8-2 Scheduled Tasks Fields 

Field
Description
Values

Directory Update

Sets the frequency and time that you want the system to automatically perform the Names Tuning and Directory Update processes.

Frequency—Daily, Weekly (Sunday - Saturday), Monthly (Day 1 - Day 31).

Time of Day—Select from the list or enter a value.

Last Completion—Shows completion time in the format MM/DD/YYYY HH:MM:SS.

Note Time of Day and Last Completion are the Speech Connect system times, and may be in a different time zone than your PC.

Backup Data

Sets the frequency and time that you want the system to automatically perform a complete system and database backup.

Frequency—Daily, Weekly (Sunday - Saturday), Monthly (Day 1 - Day 31).

Time of Day—Select from the list or enter a value.

Last Completion—Shows completion time in the format MM/DD/YYYY HH:MM:SS.

Note Time of Day and Last Completion are the Speech Connect system times, and may be in a different time zone than your PC.