曦灵

曦灵 is a Digital Human platform from Baidu AImed at teams that want virtual presenters without a full production crew. It combines avatar creation, Voice Cloning, and vIDEo or livestream delivery in one browser-based workspace. The emphASIs is on speed: short input CLIPs and plain-language Prompts are used to produce talking avatars that can read scripts, answer questions, or host a livestream.

What it does

The platform turns a topic or script into a digital human broadcast video in a few guided steps, with multilingual translation available so one script can be delivered in several languages. It also supports rAPId setup of an AI livestream room, including script rewriting, real-time audience interACTion, and product card pop-ups for commerce-oriented sessions.

Key capabilities

• 2D Portrait cloning: recreates a presenter from a short video Sample, with lip movement accuracy advertised at roughly 98.5 percent. The vendor describes a very fast turnaround for building a 1:1 digital twin.
• text-to-3D digital humans: a prompt-based route to a photorealistic 3D avatar in minutes, with conversational adjustment of APPearance, makeup, hair, and clothing.
• voice cloning: a Personal voice can be replicated from about 30 SEConds of audio, and the platform also offers curated premium voices intended for media-grade use.
• Livestream automation: script handling, audience interaction, and on-screen commerce elements are packaged for always-on virtual hosting.

Who it is for

曦灵 suits marketing and e-commerce teams producing product videos or live shopping streams, media groups needing multilingual presenter content, and corporate communications or training departments that want consistent on-camera spokespeople. It is also relevant to creators who prefer a cloned avatar over repeated filming sessions.

How to get started

Access is through a web browser with a Baidu account. A typical first project starts by choosing a template or use case, supplying a short portrait video for cloning, and either recording or cloning a voice. From there, users enter a topic or script, pick a language, and generate the video or launch a livestream room. Because the interface and documentation are primarily in Chinese, non-Chinese speakers may need translation support.

Pricing and access caveat

free trial access and paid tiers appear to be offered, but plan details, quotas, and cloning limits may vary by account and region and should be confirmed on the official site. Availability outside mainland China can be limited, and cross-border users may encounter registration or network restrictions. Feature depth, such as 3D avatar quality or premium voice licensing, may also differ between plans.