Content credentials
The high-quality models in the Azure text to speech avatar feature generate realistic avatar videos from text input. To improve the transparency of the generated content, the Azure text to speech avatar provides content credentials, a tamper-evident way to disclose the origin and history of the content. Content credentials are based on an open technical specification from the Coalition for Content Provenance and Authenticity (C2PA), a Joint Development Foundation project.
What are content credentials?
Content credentials in the Azure text to speech avatar provide customers with information about the origin of an avatar video. This information is represented by a manifest attached to the video. The manifest is cryptographically signed by a certificate that traces back to Azure text to speech avatar.
The manifest contains several key pieces of information:
| Field name | Field content |
|---|---|
"generator" |
This field has a value of "Microsoft Azure Text To Speech Avatar Service" for all applicable videos, attesting to the AI-generated nature of the video. |
"when" |
The timestamp of when the content credentials were created. |
Content credentials in the Azure text to speech avatar can help people understand when video content is generated by the Azure text to speech avatar system. For more information on how to responsibly build solutions with text to speech avatar models, visit the Text to speech transparency note.
Limitations
The content credentials are only supported in video files generated by batch synthesis of text to speech avatar, and only mp4 file format is supported.
How do I use content credentials in my solution today?
You can use content credentials by ensuring that your Azure text to speech avatar generated video files contain content credentials.
No extra set-up is necessary. Content credentials are automatically applied to all applicable videos generated by the Azure text to speech avatar.
Verifying that a video file has content credentials
You can check the content credentials of a text to speech avatar via either of the following methods:
- Content Credentials Verify webpage (contentcredentials.org/verify): This is a tool that allows users to inspect the Content Credentials of content such as a video. If an uploaded video of a text to speech avatar is created with the Speech service, the tool will show that and display the issued date and time of the Content Credentials. This screenshot shows an example avatar video upload result where the Content Credentials were issued by Microsoft.
This page shows that a video generated by Azure TTS Avatar Service has Content Credentials issued by Microsoft.
- Content Authenticity Initiative (CAI) open-source tools: The CAI provides multiple open-source tools that can be used to validate and display C2PA Content Credentials. Find the right tool for your application and get started here.