Generate talking-head videos, lip-sync replacements, and digital avatars via DashScope APIs.
npx skills add https://github.com/openakita/openakita --skill avatar-studio
Please help me install this Agent Skill. Skill: avatar-studio Source: https://github.com/openakita/openakita/tree/main/plugins/avatar-studio Command: npx skills add https://github.com/openakita/openakita --skill avatar-studio
Creating digital human videos (talking photos, lip-synced dubs, face swaps) normally requires stitching together multiple AI models, TTS engines, and billing logic by hand. This Skill orchestrates the full DashScope pipeline so an agent can produce a finished MP4 from a photo and a sentence of text.
Ask the agent to make the attached portrait photo speak the sentence "Hello, welcome to my channel" using the default voice at 480P resolution.
High-intent search queries and answers about installing and using this skill.
Use the photo_speak mode with one frontal portrait image plus either text (synthesized via CosyVoice or Edge-TTS) or an uploaded audio file. The pipeline runs wan2.2-s2v-detect face validation, then wan2.2-s2v to produce a talking-head MP4.
Use the video_relip mode, which calls DashScope videoretalk with your video plus new text or audio. Audio must be between 2 and 120 seconds, and the output is a lip-synced MP4 billed at 0.30 CNY per second.
Yes, partially. You can switch the backend to RunningHub with its own API key, or to a local ComfyUI instance which runs workflows for free. Edge-TTS also provides free speech synthesis without any paid key.
A dependency error means the input asset was rejected: the image is not a clear frontal human face, or the reference video exceeds duration limits. Replace the asset with a single-person frontal photo and keep reference videos under 30 seconds.
DashScope allows only one async task in flight per API key, so submissions are serialized through a semaphore. Additional tasks queue until the current one finishes, and task IDs expire after 24 hours on the platform.
Upload-PostUpload videos, photos, text, and LinkedIn documents to multiple social platforms via the Upload-Post API.
gooseworks-aiAutomate UGC ad creation by generating talking-head videos and combining them with product clips.
hanzoaiGenerate HeyGen avatar videos with programmatic control over avatars, voices, scripts, and backgrounds.
rongxinzyGenerate videos from text, images, or audio using the Volcengine Seedance model.
