Skip to content

Azure Speech Services.md

Latest commit

 

History

History
49 lines (39 loc) · 3.24 KB

Azure Speech Services.md

File metadata and controls

49 lines (39 loc) · 3.24 KB
Raw

Azure Speech Services

Azure Speech Services can provide cloud-hosted neural text-to-speech voices for EDDI. To use these voices, create an Azure Speech resource, then configure an Azure Speech Services provider profile in EDDI with the resource region and API key.

Create an Azure Speech resource

  1. Open https://portal.azure.com/ and sign in. If you do not have an account, create a free account to get started.
  2. Select Create a resource.
  3. Search for the Speech Azure Service and select it.
  4. Select Create.
  5. On the Basics tab
    • Choose the Azure Subscription you wish to use to create the resource. If you don't have a subscription, you can create a free account to get started.
    • Choose the Resource group to use. The resource group simply helps organize your Azure resources (don't overthink it). Create a new resource group if needed.
    • Choose a Region near your usual location. This corresponds to a regional data center - choosing a nearby region will reduce speech synthesis latency.
    • Enter a unique resource Name.
    • Choose a Pricing tier. Typically you'll be able to choose between the free tier (F0) and standard tier (S0).
  6. On the Network tab
    • (Optional) You may configure network security settings to limit access to the speech resource. Restricting access by IP address is not advised unless you know for certain that the public IP address you will be using is static.
  7. Select Review + create, correct any validation issues, then select Create. If everything is configured correctly, the resource will be deployed. This may take a few minutes.
  8. After deployment completes, select Go to resource.

Find the region and API key

  1. In the Speech resource overview, find Keys and Endpoint.
  2. Copy one API key, such as Key 1.
  3. Copy the resource Location/Region value. EDDI needs this region value along with the API key.

Keep the API key private. If a key is exposed, regenerate it in Azure and update the provider profile in EDDI.

Configure EDDI

  1. Open EDDI's Text-to-Speech tab.
  2. Select Manage Web Voice Providers....
  3. Select Add Profile.
  4. Choose Azure Speech Services as the provider type.
  5. Enter a profile name if you want a custom name.
  6. Paste the Azure resource region into Region.
  7. Paste the Azure API key into API key.
  8. Optionally enter locale filters such as en, en-GB, or de-DE to filter out voices not designed to be compatible with the speech you are generating.
  9. Select Verify to confirm EDDI can connect to the provider.
  10. Select Save.
  11. Back on the Text-to-Speech tab, select an Azure voice from the voice list and test it.

Notes

  • Azure voices require a working internet connection and an Azure Speech resource.
  • You are responsible for any costs incurred by your Azure Speech resource. The free tier (F0) provides up to 5 million characters of text-to-speech per month, which should be sufficient for typical EDDI usage.
  • Use Clear Credentials or remove the profile if you no longer want EDDI to retain the API key.
  • Local EDDI PLS lexicon files are not uploaded to Azure. EDDI can still send inline SSML pronunciation tags when they are generated by scripts or speech conversion rules.