AtBot gives you the ability to connect a Microsoft Speech service to your bots in Teams. This feature allows users to call your bots and interact with them via voice as well as text. There are a few steps to take in order to get this to work; settings in Azure, AD as well as the AtBot Admin portal. You must have configured your Bot Framework registration prior to configuring call settings.
Teams Bot Channel Settings
The first step is to enable the Teams Calling feature in AtBot and then update some settings in Azure.
- Starting in the AtBot Admin Portal, on the edit screen for your Enterprise bot, click on the Teams Call Settings tab.
- Check the box to enable this bot to take Teams calls.
- Copy the value for Webhook Endpoint.
- Navigate to your Bot Channel Registration and click on Channels > Edit the Teams channel.
- Under the Calling tab, check the box to enable calling and paste in the webhook url you copied from AtBot.
Additional Permissions and Admin Consent to Your Bot Registration
The next step is to add some additional API permissions to your bot registration as well as provide admin consent to those permissions for your organization. These steps are required for any bot framework bot that allows calling.
- On the Settings page of your Bot Channel Registration in Azure, click on the Manage link next to Microsoft App ID.
- Navigate to API Permissions and add the following Microsoft Graph Permissions as Application permissions:
- Navigate to Azure Active Directory > Enterprise Applications
- Find your bot registration and navigate to the permissions panel. There will be a large button that will allow you to Grant admin consent for your bot to your organization.
Connecting the Speech service
The final step in giving your Teams bot a voice is to create and connect to a Speech Service in your Azure tenant. Once created, grab a Key that is generated and note the region you created the service. Navigate back to the Teams Call Settings tab in AtBot.
The key that you copied from your Speech Service in Azure.
The region that your speech service is deployed to
Enter or select the voice from the list of voices by clicking the
Speech to Text Language
Select the language this bot with understand
Create a wake word file by logging in to the Custom Speech Portal
. Once you have downloaded your wake word file, click the Save Wake Word button to upload your file and your wake word. At any time, you may update just your wake word text. You may also add additional phonetic spellings of your wake word separating them with semicolons (Sophia;Sofia).
Animations for your bot must be pregenerated. Ensure that you have a bot icon and that you have saved the bot before generating animations.
Parrot Mode allows you to see exactly what is returned from the Speech to Text service. This is very helpful in understanding the different spellings you need to add to your wake word text to make your bot operate properly.