Microsoft has launched a new text-to-speech capability coupled with visual functionalities, enabling users to generate videos featuring talking avatars from text input and construct interactive bots using human images in real-time.
Known as Azure AI Speech text, currently in public preview, this feature empowers customers to fabricate synthetic videos featuring a 2D photorealistic avatar speaking.
During the 'Microsoft Ignite' event, the company highlighted that the Neural text-to-speech Avatar models are trained through deep neural networks, leveraging human video recordings as samples. The voice of the avatar is provided by a text-to-speech voice model.
The introduction of this text-to-speech avatar aims to elevate digital interactions, allowing users to craft conversational agents, virtual assistants, chatbots, and more, fostering engagement in various digital environments.
In emphasizing responsible usage, Microsoft underscores the significance of safeguarding individual and societal rights, fostering transparent human-computer interactions, and combatting the proliferation of harmful deepfakes and misleading content. Consequently, the custom avatar feature is available through limited access by registration, catering to specific use cases. Interested users can apply for access by registering their use cases.
At present, Microsoft offers two distinct text-to-speech avatar features: prebuilt and custom text-to-speech avatars.
The prebuilt text-to-speech avatars are readily available products on Azure, offering subscribers the ability to select from various avatars capable of speaking multiple languages and voices based on the text input. Customers can leverage these avatars to create video content or interactive applications, enabling real-time avatar responses for enhanced user engagement.
(With Agency Inputs)
ALSO READ | Windows 11 Update: Microsoft Expands User Freedom to Uninstall Additional Inbox Apps
ALSO READ | Microsoft rejigs top Xbox team as it sets eyes on AI and gaming
