Specify the Google Cloud Dialogflow CX speech model
The ability to configure speech models enables you to experiment with current or future speech models to use for the entire bot, or an individual turn. Configure a Google Cloud Dialogflow CX speech model for use with Genesys Cloud by creating a Custom Payload Response for your bot in the Google Cloud Dialogflow CX console.
Model selection
When you enter the custom payload response as described in the following task, you can use one of the models defined in the Model selection list in the Google Cloud Speech-to-Text guide, or you can use an empty string. If you use an empty string, Google Cloud selects the best available model. If you omit the property, then Google Cloud uses the last specified model. For example, if you set the model to “latest_short” and supply a second set of options without choosing a model, then Google Cloud uses “latest_short.”
Variant selection
When you enter the custom payload response as described in the following task, you can set the variant to one of these options:
- Standard
- Enhanced
- Empty string
If you use an empty string, then Google Cloud selects the best available variant. If you omit the property, then Google Cloud uses the last specified variant. For example, if you set the variant to “enhanced” and then supply a second set of options without choosing a variant, then Google Cloud uses “enhanced.”
Use a custom payload to specify the Google Cloud Dialogflow CX speech model
- In the Google Cloud Platform console, go to the Dialogflow CX Console.
- Select a project.
- Select an agent. The Default Start Flow for the agent appears.
- Select the page for which you want to configure the speech model.
- Select a Route.
- Under Fulfillment, click Add dialogue option and then click Custom Payload.
- To specify the Google Cloud Dialogflow CX speech model, add the following code to the custom payload response:
{ "recognition-options": { "model": "", "variant": "" } }
- Click Save.