Download and distribute boosted language model (Standalone mode)

You will be notified when the updated language model file is available. Download the ZIP file of the boosted language model, and distribute the language model to all the Speech Transcription servers that use the model.

Distributing the language model replaces the current model with the new version. The method for distributing the language model depends on whether you have the standard or real-time installations of the Speech Analytics product.

Before you begin

Send Project to the Support team
Ensure to have the email notification from the Support team containing the FTP details handy

Procedure

Download the ZIP file using the FTP details in the email notification from Customer Analytics.

The file name format is xx_YY-<model-number>-RTSA-<customer_name>-V<number>.zip, where:
- xx is the language code, for example, en for English.
- YY is the vocabulary code, for example, US or UK.
- <model-number> indicates the version of the base model, based on the number of revisions to the base language model.
- RTSA identifies that the language model is for a real-time Speech Analytics installation, and is added only for RTSA installations.
- <customer_name> is the name of the customer as defined in the project properties, for example, ACME.
- <number> is the boosted version of the language model used with the project. It number starts from 1 for a new project, and is incremented sequentially per version.
Example: en_US-5-ACME-V1.zip for standard installations, and en_US-5-RTSA-ACME-V1.zip for RTSA installations.
To distribute the language model file, do one of the following:
- For standard Speech Analytics installations, load the language model to the primary transcription server.
- For RTSA installations, distribute using the Recorder Management application.