Speech Analytics SaaS considerations

Some Speech Analytics features have specific considerations in a SaaS environment.

Considerations are indicated for each feature, and recommendations on how to approach the feature implementation in the SaaS solution.

Feature	Consideration	Recommendations
Application Servers	Every business line or language requires a dedicated Speech Application Server. Depending on the number of business lines or languages supported by the tenant, there can be more than one Application Server per tenant.	N/A
Transcription Servers	All organizations require a dedicated Speech Analytics transcription server for each language model.	Consult with Services personnel on configuring the transcription servers.
Transcription	All organizations require a unique language model generated by the Enablement Team per requested language. Automated language detection is not supported.
ICR (Intelligent Call Recording)	The system cannot be deployed with both Speech and ICR configurations. ICR is license-driven and only a single license is allowed in the system.	To achieve ICR functionality, the following privileges must be removed from the relevant roles for the appropriate tenants: Access Analytics application View Transcript
Phonetics Boosting	The Phonetics Boosting application is installed on the desktop in either connected or standalone modes. The mode determines how the language model associated with the Phonetics Boosting project is built and distributed. In connected mode, the language model is built and distributed from within the Phonetics Boosting application. In standalone mode, the language model is built offline (by Customer Support), and then distributed using the Language Model Distributor. In an MT SaaS environment, Phonetics Boosting is supported in an MT SaaS deployment using standalone mode only. Standalone mode can be used by tenants, while the distribution of the model after receiving it from Verint is done by the Service Provider (using the Language Model Distributor). Connected mode is not supported because this mode provides access to all models across the system and is not multi-tenant enabled.	Configure Phonetics Boosting for Standalone mode Build the language model and use the Language Model Distributor for distribution
Semantic Intelligence	Semantic Intelligence is supported from V15.2 HFR3 and higher. It is not supported in V15.2 HFR1 and V15.2 HFR2.	For V15.2 HFR1 and HFR2 only: Do not load an ontology into the system to ensure that Semantic Intelligence is not used.