Using Azure Speech Services

Configuration

Trucaption can be configured to use Azure Speech Services instead of the built-in browser speech recognition. To use Azure Speech Services, you will need:

• A Microsoft Azure account
• An Azure Speech Service

Once you have created a Speech Service in the Azure portal, obtain the region and a subscription key from the Overview page of the service (there will be two subscription keys, it doesn’t matter which one you choose). Then, configure Trucaption to use Azure for transcription:

1. Within the editor window, click Configure.
2. Set the transcription engine to “Azure”.
3. Enter the Location/Region from the Azure portal as Azure Region.
4. Enter the subscription key from the Azure portal as Azure Subscription Key
5. Azure Endpoint ID should be left blank unless you are using Custom Speech.

Custom Speech

Once your Azure speech account is configured, you can also choose to set up custom speech. This allows the engine to be trained to provide more accurate captions.

To set up custom speech:

1. Log in to the Azure Speech Studio.
2. Create a new project.
3. Train your custom model. (Azure Docs)
4. Deploy a custom endpoint with your trained model. (Azure Docs)
5. Obtain the Resource Key and Endpoint ID for your endpoint from Speech Studio.
6. Configure Trucaption to use your custom endpoint:
   • Set Azure Subscription Key to the Resource Key for the endpoint.
   • Set Azure Endpoint ID to the Endpoint ID.