Starting services without Management Station

 

If you are not using Management Station, you must configure and start services manually as described below.

Requirements:

• Decide which services to run on which hosts, and how many instances of each. See Deployment architecture decisions.
• Configure the number of licensing ports needed for your system. See Configuring license checkouts.
• If you are running a remote License Manager, configure its location. See Pointing products to a remote license manager.
• Configure and run each instance of each Speech Suite service separately.

Reminder: you must install voices and languages before running services. See Adding languages and voices.

Here is the basic work flow: 

1. Create one configuration file for each instance of each service on each host — For example, two recognizers require two configuration files. Typically, you create one file and make identical copies for each each instance. By convention, the copies have numbered suffixes: 01, 02, and so on. (Some services have only one instance and one configuration file.)
2. Start services in the correct order — Start Resource Manager first, and start Speech Server last, and start NTpE before NLE. (Reverse this order when stopping services.) Other services can start in any order.
3. Manage the running system — Use operating system commands to check status, stop or restart.

Note: The commands to start services include arguments to ignore the Watcher service. The watcher is only relevant when using Management Station. It is a low-cost process that can continue running even when not used.

Starting the Nuance Resource Manager

Start the Resource Manager if using Dragon Voice.

Note: For redundancy, you can run instances of the resource manager on different hosts. The first instance that starts is the primary, and the others are secondary (they can receive requests, but they relay them to the primary). If the primary fails, one of the secondary instances becomes the new primary. When the failed instance returns to service, it is secondary.

To manually configure and start the Resource Manager:

1. Ensure that ZooKeeper is configured and started. See Setting up ZooKeeper.
2. Copy the Resource Manager configuration file user-NRM.yaml, to User-nrm01.yaml.
   • Windows: get the template file from %NRM_DATA_DIR%\config\ and store in %NUANCE_DATA_DIR%\system\config\
   • Linux: get the template file from $NRM_DATA_DIR/config/ and store in $NUANCE_DATA_DIR/system/config/
3. Edit User-nrm01.yaml:
   1. If zookeeper runs on a different host, set zookeeper : servers to the hostname or IP address of that host. If using multiple ZooKeepers (recommended), specify each hostname or IP address on a separate line. Use the same port (2181) for all.
   2. Set https : host (or http : host) to the hostname or IP address of the Resource Manager host.
   3. Optional. Set the http : port and https : port listener ports. If you do this, you MUST configure all other engines to communicate on the same listener port (as described in the sections below).
4. Set the logging level on the command line or in the Resource Manager configuration file. The command-line setting takes precedence. See Diagnostic logging for valid levels. They are the same as Management Station. Logs are written to $NUANCE_DATA_DIR/system/diagnosticLogs/nrmxx.log, where xx is the instance number.
5. Start the instance.
   On Linux:

   The service runs under the user nuance and group nuance account. This command ensures that you start the instance as nuance user.

   > su -s /bin/bash nuance -c '$NRM_HOME/bin/nrm config=$NUANCE_DATA_DIR/system/config/User-nrm01.yaml config.ServiceID=0 config.LoggingLevel=STATUS watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE'

   Troubleshooting: if the Resource Manager fails to start, the cause might be related to permissions. See Resource Manager or NTpE fails to start on Linux.

   On Windows:

   > %NRM_HOME%/bin/nrm config=%NUANCE_DATA_DIR%/system/config/User-nrm01.yaml config.ServiceID=0 config.LoggingLevel=STATUS watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE

   Note: If using SNMP, set watcher.SendAlarmsToWatcher=TRUE. Nuance recommends that watcher.RestartOnFailure is always FALSE. For details, see Configuration parameters .

Starting the Natural Language Processing service

Note: Run one instance on one host. Do not start more than one instance.

Start the NLP service if using Dragon Voice recognition.

1. Copy the NLP service configuration file (user-NLPS.properties) to User-nlps01.properties.
   • Windows: get the template file from %NLPS_HOME%\config\ and store in %NUANCE_DATA_DIR%\system\config\
   • Linux: get the template file from $NLPS_HOME/config/ and store in $NUANCE_DATA_DIR/system/config/
2. Edit User-nlps01.properties:
   1. The Speech Server communicates with the NLP service. Verify these NLP service configurations: server.port = 8443 and server.ssl.enabled = true. In the configuration for every Speech Server instance in the network, ensure that server.nlps.serverAddress matches the NLP service server.port setting. A single NLP service instance can support multiple Speech Server instances.The NLP service listens on this port for Speech Server connections and messages.
   2. Set resourceManager.urls to the hostname or IP address where the Resource Manager runs. Do not use the value localhost (even if the resource manager runs on the same host as the NLP service).
   3. Set numberOfLicenses to the number of ports your license supports.
3. Optional. Set the logging level and log file location in $NLPS_HOME/config/log4j2.xml. (Set the level in the line <Logger name="com.nuance.nlps" level="STATUS" additivity="false">. For a list of valid values, see NLE and NLP service logging levels.) By default, logs are written to $NUANCE_DATA_DIR/system/diagnosticLogs/nlps.log.

Start the instance.

Note: If using SNMP, set watcher.SendAlarmsToWatcher=TRUE. Nuance recommends that watcher.RestartOnFailure is always FALSE. For details, see Configuration parameters .

Starting the Krypton service

Start the Krypton service if using Dragon Voice recognition.

Note: You can run multiple instances on one or more hosts in your network. Each instance requires a separate configuration file.

1. Copy the Krypton configuration file (krypton.yaml) to User-kryptonxx.yaml, where xx is the instance number, for example, User-krypton01.yaml, User-krypton02.yaml, User-krypton03.yaml. Do this for as many instances as you require on this host.
   • Windows: get the template file from %KR_DATA_DIR%\config\ and store in %NUANCE_DATA_DIR%\system\config\
   • Linux: get the template file from $KR_DATA_DIR/config/ and store in $NUANCE_DATA_DIR/system/config/
2. Edit each User-kryptonxx.yaml file:
   1. Increment https : port for each additional instance. Choose a standard increment, say 10. For example, if the port value = 8600 in User-krypton01.yaml, set it to 8610 in User-krypton02.yaml, 8620 in User-krypton03.yaml, and so on.
   2. Set rmRegistration : urls to the URL where the Resource Manager is running. (If you changed the Resource Manager listener port, either http : port or https : port depending on your choice of http or https protocol , you must change this URL port to the same value.) For example:

      rmRegistration:
        urls:
          - 'https://hostname_or_IP_address:9001'

   3. Optional. Set the logging level on the command line or in the Krypton configuration file. The command-line setting takes precedence. See Diagnostic logging for valid levels. Logs are written to $NUANCE_DATA_DIR/system/diagnosticLogs/kryptonxx.log, where xx is the instance number.
3. Start each instance. (Update the configuration filename for each instance.)
   On Linux:

   The service runs under the user nuance and group nuance account. This command ensures that you start the instance as nuance user.

   > su -s /bin/bash nuance -c '/usr/local/Nuance/Krypton/startEngine.sh config=$NUANCE_DATA_DIR/system/config/User-krypton01.yaml  config.ServiceID=0 config.LoggingLevel=STATUS watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE'

   If the service fails to start on or Red Hat 8, create the symbolic link described here: On Linux hosts running Red Hat Enterprise 8.

   On Windows:

   Change to %KR_HOME% and run:

   > startEngine.bat config=%NUANCE_DATA_DIR%/system/config/User-krypton01.yaml config.ServiceID=0 config.LoggingLevel=STATUS watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE

   Note: If using SNMP, set watcher.SendAlarmsToWatcher=TRUE. Nuance recommends that watcher.RestartOnFailure is always FALSE. For details, see Configuration parameters .

Starting the Nuance Text Processing Engine (NTpE) service

Start the NTpE service if using Dragon Voice recognition with semantic interpretation. (The service is not needed for Krypton-only recognition.)

Note: You can run multiple instances on one or more hosts in your network. Each instance requires a separate configuration file.

1. Copy the NTpE configuration file (user-NTPE.yaml) to User-ntpexx.yaml, where xx is the instance number, for example, User-ntpe01.yaml, User-ntpe02.yaml, User-ntpe03.yaml. Do this for as many instances as you require on this host.
   • Windows: get the template file from %TEXTPROC_HOME%\config\ and store in %NUANCE_DATA_DIR%\system\config\
   • Linux: get the template file from $TEXTPROC_HOME/config/ and store in $NUANCE_DATA_DIR/system/config/
2. Edit each User-ntpexx.yaml file:
   1. Increment the port: parameter for each additional instance. Choose a standard increment, say 10. For example, if the port value = 9092 for User-ntpe01.yaml, set it to 9102 in User-ntpe02.yaml, 9112 in User-ntpe03.yaml, and so on.
   2. Set resourceManager : enabled to true, and set resourceManager : urls to the host where the Resource Manager runs (even if the Resource Manager runs on the localhost). For example, https://hostname_or_ip_address:9001. Do not use the default setting localhost or the loopback address 127.0.0.1. (If you changed the Resource Manager listener port, either http : port or https : port depending on your choice of http or https protocol, you must change this URL port to the same value.)
   3. Ensure resourceManager : ntpeHost is set to the hostname or IP address of this host. Do not use the default setting localhost or 127.0.0.1.
   4. Optional. Set the logging level on the command line or in the NTpE configuration file. The command-line setting takes precedence. See Diagnostic logging for valid levels. They are the same as Management Station. Logs are written to $TEXTPROC_HOME/logs/service_portnum-x, where portnum is the port for that instance and x is the instance number, for example, service_9092-1.log, service_9102-2.log, and so on.

      Optional. You can change the log filename and location in the NTpE configuration file.

3. Start each instance. (Update the configuration file name for each instance.)
   On Linux:

   The service runs under the user nuance and group nuance account. This command ensures that you start the instance as nuance user.

   Change to $TEXTPROC_HOME and run:

   > su -s /bin/bash nuance -c 'NTPE_ID=1 ./startEngine.sh config=$NUANCE_DATA_DIR/system/config/User-ntpe01.yaml config.ServiceID=0 config.LoggingLevel=STATUS watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE'

   If the service fails to start on or Red Hat 8, create the symbolic link described here: On Linux hosts running Red Hat Enterprise 8.

   On Windows:

   Change to %TEXTPROC_HOME% and run:

   > startEngine.bat 1 config=%NUANCE_DATA_DIR%/system/config/User-ntpe01.yaml config.ServiceID=0 config.LoggingLevel=STATUS watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE

   Note: If using SNMP, set watcher.SendAlarmsToWatcher=TRUE. Nuance recommends that watcher.RestartOnFailure is always FALSE. For details, see Configuration parameters .

Starting the Natural Language Engine (NLE) service

Start the NLE service if using Dragon Voice recognition with semantic interpretation. (The service is not needed for Krypton-only recognition.)

Note: Do not run more than one instance on a single host. To start instances on other hosts, just repeat the steps using the same port number in the command line for each.

1. Copy the NLE configuration file (nle.properties) to User-nle01.properties.
   • Windows: get the template file from %NLE_HOME%\config\ and store in %NUANCE_DATA_DIR%\system\config\
   • Linux: get the template file from $NLE_HOME/config/ and store in $NUANCE_DATA_DIR/system/config/
2. Edit User-nle01.properties:
   1. Ensure nle.host is set to the hostname or IP address of this host. Do not use the default setting localhost.
   2. Register NLE with the Resource Manager:

      Set rm.enabled to true.

      Set rm.uris to the URI of that host (even if the Resource Manager runs on the localhost). For example, https://hostname_or_ip_address:9001. (If you changed the Resource Manager listener port, either http : port or https : port depending on your choice of http or https protocol, you must change this URL port to the same value. Use the same port for all NLE services.) 

      Set ntpe.resourceManagerEnabled to true. (This tells NLE to "talk" to the Resource Manager to find a suitable NTpE instance.)

   3. If NTpE is running on a host that is not the NLE host, set ntpe.uri to the URI of that host. For example, wss://hostname_or_ip_address:9092/NTpE. (The service ignores this parameter if you use the Resource Manager.)
3. Set the logging level on the command line. See Diagnostic logging for valid levels.
4. Optional. Set the location for log files. By default, logs are written to $NUANCE_DATA_DIR/system/diagnosticLogs/nle-portnum.log, where portnum is the port for that instance. (For example, nle_9091.log, nle-9101.log, and so on.) To change the location, set the logPath parameter in $NLE_HOME/config/log4j2.yaml.

Start the instance.

Note: If using SNMP, set watcher.SendAlarmsToWatcher=TRUE. Nuance recommends that watcher.RestartOnFailure is always FALSE. For details, see Configuration parameters .

Starting the Nuance recognition service

You must start at least one instance of the recognition service in your network. (You must run the Nuance recognition service when using Dragon Voice even if you don't use Nuance Recognizer.)

Note: You can run multiple instances on one or more hosts in your network. Each instance requires a separate configuration file.

1. Copy the Recognizer configuration file (user-NRS.xml) to User-nrsxx.xml, where xx is the instance number, for example, User-nrs01.xml, User-nrs02.xml, User-nrs03.xml. Do this for as many instances as you require on this host. Get the configuration file here: 
   • Windows: get the template file from install_path\Recognizer\Recognizer Service\config\ and store in %NUANCE_DATA_DIR%\system\config\
   • Linux: get the template file from install_path/Recognizer_Service/config and store in $NUANCE_DATA_DIR/system/config/
2. Edit each User-nrsxx.xml file and set swirec_license_ports and swiep_license_ports to the number of ports your license supports. Configure other parameters as needed. For additional details, see Configuration without the Management Station
3. Optional. Set logging levels on specific subcomponents of the recognition service, enable or disable tags in $SWISRSDK/config/defaultTagmap.txt. See Configuring Recognizer diagnostic logging. This sets logging for all service instances. To control logging for a particular instance, add relevant tags to Recognizer configuration file. Typically, you set logging levels only if requested by Technical Support. Logs are written to $NUANCE_DATA_DIR/system/diagnosticLogs/nrsx.log, where x is the instance number.

4. On Linux, change to install_path/Recognizer and run the command >. ./SETUP-env.sh to source environment variables.

   (Note: The command syntax has <dot><space><dot> where the first dot sources the script, and the second dot specifies its location.) 

5. Start each instance. (Update the configuration file name and port number for each instance. The port value is typically incremented by 10.)
   On Linux:

   The service runs under the user nuance and group nuance account. This command ensures that you start the instance as nuance user.

   > su -s /bin/bash nuance -c 'nuance-server -servlet nrs -port 8200
   -DDiagFileName=$NUANCE_DATA_DIR/system/diagnosticLogs/nrs1.log -DSWIUSERCFG=$NUANCE_DATA_DIR/system/config/User-nrs01.xml  config.ServiceID=0 watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE'

   On Linux as a daemon:

   Note: Only use this start up mechanism if running one instance of the service.

   If you have one recognition service, you can start it as a Linux daemon. An advantage is that the service restarts automatically after a reboot.

   To start the service as a daemon, change to the install_path/Recognizer_Service/amd64/bin and run the NRSservice.sh command.

   Command format:

   NRSservice.sh -servlet nrs [-port number] [-config myConfigFile] 
   [-DDiagFileName=myPath] [-args…] [-u]

   Argument	Description
   -servlet nrs	Required syntax.
   -port	Optional. An available port on the host. Nuance recommends port 8200. The default is 8200
   -config	Optional. The path to the configuration file. The default is $NUANCE_DATA_DIR/system/config/User-nrs01.xml
   -u	Optional. Uninstalls the daemon (all other arguments will be ignored).

   To control the daemon, use these command formats:

   On CentOS 7 and above or Red Hat 7 and above:

   systemctl [stop|start|restart|status] NRSservice

   Uninstall:

   To uninstall the daemon, use this command:

   NRSservice.sh -u

   Note: Uninstall the daemon BEFORE uninstalling Nuance Recognizer.

   On Windows:

   > nuance-server -servlet nrs -port 8200 -DDiagFileName=%NUANCE_DATA_DIR%/system/diagnosticLogs/nrs1.log -DSWIUSERCFG=%NUANCE_DATA_DIR%/system/config/User-nrs01.xml config.ServiceID=0  watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE

   Note: If using SNMP, set watcher.SendAlarmsToWatcher=TRUE. Nuance recommends that watcher.RestartOnFailure is always FALSE. For details, see Configuration parameters .

Starting the Nuance vocalizer service

Start the vocalizer service if using text-to-speech.

Note: You can run multiple instances on one or more hosts in your network. Each instance requires a separate configuration file.

1. Copy the Vocalizer configuration file (user-NVS.xml) to User-nvsxx.xml, where xx is the instance number, for example, User-nvs01.xml, User-nvs02.xml, User-nvs03.xml. Do this for as many instances as you require on this host. Get the configuration file here: 
   • Windows: get the template file from install_path\Vocalizer for Enterprise\NVS Service\config\ and store in %NUANCE_DATA_DIR%\system\config\
   • Linux: get the template file from install_path/NVS_Service/config/ and store in $NUANCE_DATA_DIR/system/config/
2. Edit each User-nvsxx.xml file:
   1. Set the tts_license_ports parameters to the number of ports allowed by your license.
   2. In the <cache_directory> parameter, increment NVSxx for the instance number.
   3. Add <log_file_base_name>${TMPDIR}/nvn_log_${USER}-NVSxx_</log_file_base_name> substituting xx with the instance number.
   4. Add <event_log_file_base_name>${TMPDIR}/nvn_event_log_${USER}-NVSxx_</event_log_file_base_name> substituting xx with the instance number.
   5. Add <event_log_input_text_file_base_name>${TMPDIR}/nvn_event_log_input_text_${USER}-NVSxx_</event_log_input_text_file_base_name> substituting xx with the instance number.
3. To set the logging level, use log_level. Typically, you would do this only if requested by Technical Support. Logs are written to $NUANCE_DATA_DIR/system/diagnosticLogs/nvs-componentx.log, where component is client or server., and x is the instance, for example, nvs-client1.log.
4. On Linux only: Add the Nuance Recognizer lib path to the LD_LIBRARY_PATH environment variable.

   > export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:install_path/Recognizer/amd64/lib/

5. Start each instance. (Update the configuration file name and port number for each instance. The port value is typically incremented by 10.)
   On Linux:

   The service runs under the user nuance and group nuance account. This command ensures that you start the instance as nuance user.

   > su -s /bin/bash nuance -c 'nuance-server -servlet nvs -port 9200 
   -DDiagFileName=$NUANCE_DATA_DIR/system/diagnosticLogs/nvs-server1.log nserver.nvs.VocalizerUserCfgFile=$NUANCE_DATA_DIR/system/config/User-nvs01.xml config.ServiceID=0 watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE'

   On Linux as a daemon:

   Note: Only use this start up mechanism if running one instance of the service.

   If you have one Vocalizer service, you can start it as a Linux daemon. An advantage is that the service restarts automatically after a reboot.

   To start the service as a daemon, change to the install_path/NVS_Service/amd64/bin and run the NVSservice.sh command.

   Command format:

   NVSservice.sh -servlet nvs [-port number] [-config myConfigFile] [-args…] [-u]

   Argument	Description
   -servlet nvs	Required syntax.
   -port	Optional. An available port on the host. Nuance recommends port 9200. The default is 9200
   -config	Optional. The path to the configuration file. The default is $NUANCE_DATA_DIR/system/config/User-nvs01.xml
   -u	Optional. Uninstalls the daemon (all other arguments will be ignored).

   To control the daemon, use these command formats:

   On CentOS 7 and above or Red Hat 7 and above:

   systemctl [stop|start|restart|status] NVSservice

   Uninstall:

   To uninstall the daemon, use this command:

   NVSservice.sh -u

   Note: Uninstall the daemon BEFORE uninstalling Vocalizer.

   On Windows:

   > nuance-server -servlet nvs -port 9200 -DDiagFileName=%NUANCE_DATA_DIR%/system/diagnosticLogs/nvs-server1.log nserver.nvs.VocalizerUserCfgFile=%NUANCE_DATA_DIR%/system/config/User-nvs01.xml config.ServiceID=0 watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE

   Note: If using SNMP, set watcher.SendAlarmsToWatcher=TRUE. Nuance recommends that watcher.RestartOnFailure is always FALSE. For details, see Configuration parameters .

Starting Nuance Speech Server

You must start at least one instance of the Speech Server in your network.

Note: You can run multiple instances on one or more hosts in your network. Each instance requires a separate configuration file.

1. Copy the NSS configuration file (user-NSS.cfg) to User-nssxx.txt, where xx is the instance number, for example, User-nss01.txt, User-nss02.txt, User-nss03.txt. Do this for as many instances as you require on this host. Get the configuration file here: 
   • Windows: get the template file from install_path\Speech Server\server\config\ and store in %NUANCE_DATA_DIR%\system\config\
   • Linux: get the template file from install_path/Speech_Server/server/config/ and store in $NUANCE_DATA_DIR/system/config/
2. Edit each NSS configuration file:
   1. Update (and add) the parameters listed below. Values in bold get updated for each instance. These are a subset of parameters in NSSserver.cfg so add and configure other parameters as needed. For example, if you need to add any MRCPv1 parameters, get them from NSSserver.cfg and copy them here. These show settings in the first copy, User-nss01.txt.
      Notes: For port or Port settings, choose a standard increment, say 10. For example, if server.mrcp2.sip.transport.tls.port is set to 6076 in User-nss01.txt, set it to 6086 in User-nss02.txt, 6096 in User-nss03.txt, and so on.If these services are co-located with Speech Server, you can keep the default setting of localhost. Otherwise, change the server.* address parameters to the hostname or IP address of the host (s) running these services. Since a specific Speech Server communicates with a specific recognition service and a specific vocalizer service, you need to update the port values as well, if running more than one instance of each.

      When setting the values below, mirror the correct syntax in NSS configuration file. The format is parameter VXIString value. (If a property is not already in the configuration file, you need to add it.) 

      (update these properties for each Speech Server instance)
      Property	Type	Value
      server.log.filename	VXIString	$NUANCE_DATA_DIR/system/diagnosticLogs/nss1.log
      server.mrcp2.sip.transport.udp.port	VXIInteger	5060 (change from 5066 in user-NSS.txt)
      server.mrcp2.sip.transport.tcp.port	VXIInteger	5060 (change from 5066 in user-NSS.txt)
      server.mrcp2.sip.transport.tls.port	VXIInteger	5061 (change from 5067 in user-NSS.txt)
      server.mrcp2.transport.tcp.port	VXIInteger	6075
      server.mrcp2.transport.tls.port	VXIInteger	6076
      server.rtp.port	VXIInteger	7892
      server.nvs.Address	VXIString	localhost:9200 (uncomment if needed)
      server.nvs-client.Port	VXIInteger	9305
      server.nvs-client.LogPath	VXIString	$NUANCE_DATA_DIR/system/diagnosticLogs/nvs-client1.log
      server.mrcp1.transport.port	VXIInteger	4900
      server.nrs.serverAddress	VXIString	localhost:8200 (uncomment if needed)
      server.mrcp2.osrspeechrecog.userconfig.path	VXIString	$NUANCE_DATA_DIR/system/config/User-nrs01.xml (uncomment if needed)
      server.callLog.tcp.port	VXIInteger	10101
      server.callLog.tls.port	VXIInteger	10102

      (add these properties to each Speech Server instance)
      Property	Type	Value
      server.mrcp1.osrspeechrecog.userconfig.path	VXIString	$SWISRSDK/confg/Baseline.xml (uncomment if needed)
      server.mrcp1.resource.2.cfgprefix	VXIString	server.mrcp1.rsspeechsynth. (the value ends with a dot)
      server.mrcp1.resource.2.dll	VXIString	mrcp1nvsspeechsynth
      server.mrcp1.resource.2.name	VXIString	Nuance Vocalizer
      server.mrcp1.resource.2.url	VXIString	media/speechsynthesizer
      server.mrcp1.resource.3.cfgprefix	VXIString	server.mrcp1.osrspeechrecog. (the value ends with a dot)
      server.mrcp1.resource.3.dll	VXIString	mrcp1osrspeechrecog
      server.mrcp1.resource.3.name	VXIString	Nuance Open Speech Recognizer for MRCPv1
      server.mrcp1.resource.3.url	VXIString	media/speechrecognizer
      server.mrcp2.osrspeechrecog.userconfigTemplate.Path	VXIString	$SWISRSDK/confg/Baseline.xml (uncomment if needed)
      server.mrcp2.resource.0.cfgprefix	VXIString	server.mrcp2.osrspeechrecog. (the value ends with a dot)
      server.mrcp2.resource.0.dll	VXIString	nrsspeechrecog
      server.mrcp2.resource.0.name	VXIString	Nuance Open Speech Recognizer
      server.mrcp2.resource.0.type	VXIString	speechrecog   dtmfrecog
      server.mrcp2.resource.1.cfgprefix	VXIString	server.mrcp2.rsspeechsynth. (the value ends with a dot)
      server.mrcp2.resource.1.dll	VXIString	nvsspeechsynth
      server.mrcp2.resource.1.name	VXIString	Nuance Vocalizer
      server.mrcp2.resource.1.type	VXIString	speechsynth
      server.mrcp2.resource.4.cfgprefix	VXIString	server.mrcp2.recorder. (the value ends with a dot)
      server.mrcp2.resource.4.dll	VXIString	recorder
      server.mrcp2.resource.4.name	VXIString	Nuance Speech Recorder
      server.mrcp2.resource.4.type	VXIString	recorder
      server.mrcp2.sip.maxCountOfSession	VXIInteger	96 (modify to an appropriate value for your deployment)

   2. If using Dragon Voice recognition, connect each Speech Server instance to a single instance of the Natural Language Processing service: add server.nlps.serverAddress to each User-nssxx.txt file and set it to wss://localhost:8443/nlps. (If the NLP service is not running on the same host as Speech Server, change localhost to the the correct hostname or IP address.) For example:

      server.nlps.serverAddress VXIString wss://localhost:8443/nlps

      (Alternatively, you can set the serverAddress property in each User-nssxx.txt file. By default, the property is a comment in the configuration files.)

3. Optional. To set logging levels on specific subcomponents of Speech Server, enable or disable relevant tags using server.log.diagTag.xxxx. Typically, this is set only when requested by Technical Support. Logs are written to $NUANCE_DATA_DIR/system/diagnosticLogs/nssx.log, where x is the instance number.
4. On Linux, you must source environment variables. Change to install_path/Speech_Server/server and run this command:

   > . ./SETUP-env.sh

   (Note: The command syntax has <dot><space><dot> where the first dot sources the script, and the second dot specifies its location.) 

5. Start each instance. (Update the configuration and log filenames for each instance.)
   On Linux:

   The service runs under the user nuance and group nuance account. This command ensures that you start the instance as nuance user.

   > su -s /bin/bash nuance -c '$NSSSVRSDK/bin/NSSserver -u $NUANCE_DATA_DIR/system/config/User-nss01.txt -DDiagFileName=$NUANCE_DATA_DIR/system/diagnosticLogs/nep1.log config.ServiceID=0 watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE'

   On Linux as a daemon:
   

   Note: Only use this start up mechanism if running one instance of the service.

   If you have one Speech Server service, you can start it as a Linux daemon. An advantage is that the service restarts automatically after a reboot.

   Prepare in advance:

   • If /etc/systemd/system/NSSservice.service already exists:

     As root, insert config.ServiceID=0 into the "ExecStart" command:

     ExecStart=/usr/bin/bash -c '. /usr/local/Nuance/Speech_Server/server/SETUP-env.sh; /usr/local/Nuance/Speech_Server/server/bin/NSSserver -d -config /usr/local/Nuance/Speech_Server/server/config/NSSserver.cfg -u /var/local/Nuance/system/config/User-nss01.txt config.ServiceID=0 watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE > /dev/null 2>&1 ; pgrep NSSserver > /tmp/NSSservice.pid'

   • If /etc/systemd/system/NSSservice.service does not exist:

     As root, edit the file /usr/local/Nuance/Speech_Server/server/bin/NSSservice.service and insert config.ServiceID=0 into the "ExecStart" command:

     ExecStart=/usr/bin/bash -c '. <NSSSVRSDK>/SETUP-env.sh; <NSSSVRSDK>/bin/NSSserver -d -config <NSSSVRSDK>/config/NSSserver.cfg <NSSARGUMENTS> config.ServiceID=0 watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE > /dev/null 2>&1 ; pgrep NSSserver > /tmp/NSSservice.pid'

   To start the service as a daemon, change to the install_path/Speech_Server/server/bin and run the NSSservice.sh command.

   Command format:

   NSSservice.sh [-config myConfigFile] [-u]

   Argument	Description
   -config	Optional. The path to the configuration file. The default is $NUANCE_DATA_DIR/system/config/User-nss01.cfg
   -u	Optional. Uninstalls the daemon (all other arguments will be ignored).

   To control the daemon, use these command formats:

   On CentOS 7 and above or Red Hat 7 and above:

   systemctl [stop|start|restart|status] NSSservice

   Uninstall:

   To uninstall the daemon, use this command:

   NSSservice.sh -u

   Note: Uninstall the daemon BEFORE uninstalling Nuance Speech Server.

   On Windows:

   Change to %NSSSVRSDK%/bin and run:

   > nssserver.exe -u %NUANCE_DATA_DIR%\system\config\User-nss01.txt 
   -DDiagFileName=%NUANCE_DATA_DIR%\system\diagnosticLogs\nep1.log config.ServiceID=0 watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE

   On Windows as a service:

   The Speech Suite installer creates a disabled Windows service for Nuance Speech Server with default startup values.

   To run it:

   1. As Administrator using regedit, navigate the following registry key:

      HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Services\NuanceSpeechServer

   2. Under value "ImagePath", modify the path to include "config.ServiceID=0" as below:

      "C:\Program Files (x86)\Nuance\Speech Server\Server\bin\NSSserver.exe" -u C:\ProgramData\Nuance\Enterprise\system\config\User-nss01.txt -DDiagFileName=C:\ProgramData\Nuance\Enterprise\system\diagnosticLogs\nep1.log -s start config.ServiceID=0 watcher.RestartOnFailure=FALSE watcher.SendAlarmsToWatcher=FALSE

   3. Use the Windows Services administrative tool to change the "Nuance Speech Server" service to automatic delayed. Doing this ensures the Speech Server will start after a reboot.

   Note: If using SNMP, set watcher.SendAlarmsToWatcher=TRUE. Nuance recommends that watcher.RestartOnFailure is always FALSE. For details, see Configuration parameters .