Configure language model in server roles

Configure a tenant-specific language model for the new tenant and create a separate Speech Transcription Server cluster.

Before you begin 

Verify that you have installed and configured a new Speech project (see the Enterprise Manager Configuration and Administration Guide)

Procedure 

1. Go to System Management. Under Enterprise, select Settings.

2. Define a different tenant-specific language model, and create a separate Speech Transcription Server cluster. Associate the new tenant with that cluster.

3. Select the language and vocabulary for the Speech Application Server:

   1. Select the Speech Application Service server role.

   2. Expand the Analytics Application Server Language settings and select the Language and Vocabulary.

   The vocabulary must have either the term Custom or the name of the tenant for whom you are setting the language model, indicating that it is tenant-specific.

   Examples:

   Language = English and Vocabulary = United States Acme

   Language = English and Vocabulary = United States Custom

   Language = English and Vocabulary = United States (not tenant-specific)

4. Select the language and vocabulary for the Speech Transcription Server:

   1. Select the Speech Transcription Service server role.

   2. Expand the Transcription Language settings, and select the same Language and Vocabulary you selected in step 3.

5. Restart the Speech Application Service and the Speech Transcription Service server roles.

6. If the tenant is moving from a shared to a tenant-specific model, or from a tenant-specific to a shared model, do the following:

   1. Go to Interactions. Under Administration, select Project Rules Manager.

   2. Select the relevant rule, and in Transcribe Parameters, change the Transcription Vocabulary as required, to either the tenant-specific or the shared model.

Related information 

Speech Transcription Server Cluster Considerations (Enterprise Manager Configuration and Administration Guide)