Provide a 3-minute video to quickly generate an Avatar identical to the real person, with facial features, movements, and expressions fully mimicking the original. You only need to input text or audio to produce an Avatar broadcasting video quickly.
Note:
The final customized Avatar will closely match the makeup, skin tone, facial expressions, and movements of the person in the provided video. The original lighting from the video will also be retained. If you require beautification or brightness adjustments, you can enhance the original video before uploading it. Ensure that the video is edited to your satisfaction before submission.
Process Guide
1. Prepare video/audio material.
Avatar customization: Quickly generate an Avatar identical to the real person, with facial features, movements, and expressions fully mimicking the original.
Use an existing video with spoken content of at least 1 minute (unedited version).
Note
Video quality: The face should be clear and not blurry, with well-defined edges even when zoomed in. The video should be stable with no shaking.
Model performance: The eyes should be looking directly at the camera, with no significant head turns or tilts, and the face should remain unobstructed throughout the video.
Filming key points: The video should begin with 1-3 seconds of silence with the mouth closed. The entire video must be unedited, with no frame skips, and the total length must exceed 3 minutes.
Environmental voice: No other voices should be mixed in, and there should be no significant background noise. The audio and video must be synchronized, with the voice matching the lip movements.
This method does not support background replacement; the background will remain consistent with that of the original video.
Voice clone: Quickly generate a timbre that matches the original speaker's voice.
Record 100 sentences of audio (usually 15-20 minutes), recommended format is wav for lossless compression.
Voice clone is currently not available as a standalone service; it must be paired with Avatar customization for a complete solution.
2. Prepare a personal Avatar customization authorization letter.
To protect the legal rights of customers and the individual being customized, according to the Provisions on the Administration of Deep Synthesis of Internet-Based Information Services, appropriate measures need to be taken to ensure that legal authorization and consent from the user have been obtained. After the user authorization is obtained, the customer can provide the authorization materials to Tencent Cloud through online APIs or offline methods. Each Avatar requires the submission of the individual's authorization materials, and two authorization modes are supported: video-verbal authorization and written authorization (preferably both authorization modes should be provided).
Before submission, it is necessary to perform a self-check on the material. Once the custom material passes the self-check, it can be submitted for training through the API interface or the Digital Human platform.
4. Confirm the customization effect and service usage.
You can check the customization progress of the Avatar through Avatar Image Customization and Voice Clone API Documentation provided in the API documentation. Service usage can be accessed through the broadcasting or live streaming aPaaS APIs.
Yes
No
Was this page helpful?