As a way to fine-tune Agent Assist’s article recommendations and the speech-to-text (STT) functionality, you can set up a language selection step in Studio that will display the transcription and recommended content in the desired language.
By setting up this step in Studio, you can select specific languages and different STT services, as well as the situations in which these configurations apply - for example, to define whether the same language is used by all the agents, or only by a specific ring group or number.
To configure the STT service’s behavior, you need to edit the Studio flow where you want to change the language. At the start of a call, it will be read by Agent Assist, and the appropriate STT service and model will be automatically selected before the call is routed to an agent.
To add a language selection step in a Studio flow, proceed as follows:
1. Select the Studio flow and add a Conditional Statement component after the initial step, to configure the STT service used by Agent Assist to send audio and receive transcriptions. We’ve named ours “STT Service.
2. In the Conditional Statement’s Preferences tab, create a variable and name it “stt_service”.
3. Open the Exits tab and set up the following condition:
stt_service Equals Undefined
4. Provide a name for the exit, which must correspond to the required STT service and language code. In our example, we’ve used “S02”, because we intend to configure the STT service in English (en-GB). Please refer to the table below to find the correspondence between the code and the desired language:
5. In “Send to”, add a new Conditional Statement component, that will be used to define the language of the audio processing. We’ve named ours “STT language model”.
6. On the Preferences tab of the new Conditional Statement component, add the variable “stt_model”.
7. Open the Exits tab and set the condition to the new variable:
stt_model Equals Undefined
8. In “Exit name”, specify the language model available for the corresponding STT code chosen in Step 4. We’ve selected “en-GB”.
9. In “Send to”, add the next component to your Studio flow.
10. Click on the Manage Context button and check all of the newly created variables. Click Save to complete the process.
11. Save and publish your flow. All calls that go through this Studio flow will then be transcribed in the specified language and surface knowledge that is relevant to the interaction.
Note: Remember to assign the flow to a phone number so that all calls to that number use the flow to which the STT service and language model variables have been added.
For further information on how to configure a Studio flow, please refer to our Studio Advanced Documentation.