DigitGPTAvatars, Voices & Personas
Core Studio WorkflowA digital human combines a visual avatar, a vocal profile, a conversational language model and role-specific instructions. Configure each layer independently so the assistant reflects your brand and use case.
Core persona settings
| Setting | Best practice |
|---|---|
| Avatar | Use the studio preview to evaluate visual presence and framing for your audience. |
| Voice | Listen to voice samples and test your target languages, brand terms and numbers. |
| Language model | Select from the models enabled in your studio environment based on reasoning depth and speed. |
| Instructions | Define the role, approved knowledge scope, conversational tone and fallback behavior when information is unavailable. |
| Greeting | Open with a clear disclosure of the AI role and one focused question to guide the visitor. |
Multilingual conversations
The speech recognition engine understands 70+ languages out of the box, and visitors can also interact via keyboard at any time. In public demos or custom deployments, the assistant can switch conversational languages on request when instructed to do so.
Because speech recognition and vocal synthesis operate as distinct layers, vocal fluency and accent depend on the specific voice selected in the studio. For multilingual deployments, specify the supported languages in your assistant’s instructions and test greetings, numbers and domain terminology in each target language.
Custom avatars
Workspace owners and administrators can upload custom portrait assets where enabled for the workspace. The studio uploader accepts JPEG, PNG or WebP portraits up to 4 MB. Only upload likenesses for which you hold full commercial rights and explicit consent.