Worldwide [change] |Account |Register |About Cisco
Guest

Hierarchical Navigation

Cisco Unity

Speech Connect: "No More Names Present" Message Occurs

Downloads

Document ID: 109546


Contents

Introduction
Prerequisites
      Requirements
      Components Used
      Conventions
Problem
Solution
Related Information

Introduction

Speech Connect is an advanced, speech-enabled automated attendant system for the enterprise. Speech Connect allows customers, employees, and management to find each other without the need to know multiple phone numbers. On the phone, callers say names, departments, products, or services that are associated with an extension, and Speech Connect transfers the call.

This document discusses how to troubleshoot the speech recognition issue with the Speech Connect .

Prerequisites

Requirements

Cisco recommends that you have knowledge of this topic:

Components Used

The information in this document is based on the Speech Connect 1.0.2 and earlier.

The information in this document was created from the devices in a specific lab environment. All of the devices used in this document started with a cleared (default) configuration. If your network is live, make sure that you understand the potential impact of any command.

Conventions

Refer to Cisco Technical Tips Conventions for more information on document conventions.

Problem

Sometimes, speech recognition does not appear to work at all and callers ask for people in the directory hear these messages:

No more names to present

OR

Sorry, I did not understand.

Results (if any) that are presented to the caller are the lower scoring speech recognition results and are most likely incorrect.

The $NUANCE/session directory is the normal location for recordings made during a call. Once the directory is full, new recordings of the callers utterances cannot be made. This disrupts the processing of speech recognition results and although Speech Connect works fine internally, it does not present the caller with the correct results. The recordings (.wav files) are also noted in log files that detail each speech recognition request. This issue can be verified if you look at the logs found in /opt/cisco/speechconnect/logs/asr/{date}. The file /opt/cisco/speechconnect/logs/asr/{date}/{date}_asr.log contains a Master record for each name request and a detail record for each hypothesis returned from recognition processing. An example of a correct entry in the asr log file appears here.

Note: You can see the "master" record with a *.wav file listed, and with names returned in the detail record.

M |2008-12-1
	 9:18:27|lxsc1.lyrix.com|Company|1881|GETPERSON|CR
	 |3|3|3|http://localhost/EJDDPCCNAAAPDJICAAAAAAAA/20081201091822_019176.wav|"Peter
	 Parker" < sip:3076@10.0.112.247> D|1|1|N|-|55|harry osborn|2|15
	 15|102 D|2|2|N|-|12|flash thompson|2|12 12|121

When $NUANCE/session directory is full, the .wav file cannot be saved and the log file appears as such.

Note: The *.wav file is missing from both master records in this example. For the first record, since there are no details to present to the user, Speech Connect asks the caller to repeat their request. For the second Master record, "robert smith" is presented to the caller. But, the best result, which is not shown due to the issue, is missing and subsequently not presented to the caller.

M |2008-11-21
	 13:48:08|mycompany-sys1|Company|32073|GETPERSON|C- |1|1|1|| M |2008-11-21
	 13:48:12|mycompany-sys1|Company|32073|GETPERSON|C- |1|1|1|| D|1|1|-|-|6|robert
	 smith|2|6 6|49D72A83-9DB4-455D-B1A0-3D81285D2024

A large number of Master records without corresponding detail records is an indication of an issue. This behavior is a result of the $NUANCE/session directory that contains too many items. The directories and files $NUANCE/session need to be removed in order to return the system to normal functioning. The directory $NUANCE/session (/usr/local/Nuance/SpeechServer_5/server/session) contains more than 32,000 files.

You can confirm the number of entries in the directory with this command:

ls $NUANCE/session |wc -l

You can also review the current asr log file with this command:

more /opt/cisco/speechconnect/logs/asr/{date}/{date}_asr.log

where {date} is in yyyymmdd format.

Solution

This issue has been fixed in Speech Connect version 1.0.3. Upgrade to Speech Connect version 1.0.3 in order to resolve this issue. Once this is done, the files are purged during the standard overnight process.

Once the upgrade is performed, then you can also have the files purged if you log into the Speech Connect system and run the purge command.

Log into the system as scadmin and change to root: su ? (enter the root password). Run this command:

/opt/cisco/speechconnect/bin/SEAA_PurgeSessionData.sh

Hit Enter after you enter the previous command. This should purge the old data that causes the problem, and the system should be fine going forth.

Note: You can determine what time the nightly process is scheduled to run if you use the web interface: Config > Scheduled Tasks > Directory Update and note the time in the Time of Day column, which shows the time the next nightly process runs.

Related Information

Updated: Feb 25, 2009 Document ID: 109546