synthesia-avatar-video

What it does

Produces presenter-led AI avatar videos using Synthesia Studio and Synthesia API. Applies to business, training, product, or localization videos where controlled avatar performance and brand consistency are important. Covers planning, avatar/voice/language selection, scripted scenes, template personalization, API job lifecycle, localization/dubbing, consent/safety checks, artifact custody, troubleshooting, and final QA.

How it works

• Route the work for Synthesia Studio manually edited videos or API-based outputs: from scratch via POST /v2/videos, or from templates via POST /v2/videos/fromTemplate.
• Include personalized batches by varying variables (names, roles, product details, CTAs, languages, avatars).
• Make decisions on avatar types and consent (Custom Avatar, Personal Avatar, Studio Avatar, stock avatar) and ensure rights/safety checks before generation.
• Use defined API inputs: for API-from-scratch, provide input with avatar and background, plus avatarSettings, scriptText or scriptAudio, and soundSettings as needed.
• If using templates, supply templateId and templateData, optionally brandKitId.
• Manage job lifecycle: obtain the video id, poll status via GET /v2/videos/{video_id}, treat complete as ready, in_progress/approved not final, and error/rejected for triage. Download private links promptly and store final MP4 and metadata in the artifact system.
• Maintain a custody manifest with details like provider, mode, video_id, template_id, avatar_ids, voice_ids, languages, visibility, and status_history.

When to use it

• For Synthesia Studio videos created from scenes, scripts, templates, avatars, brand kits, media, captions, and translations.
• For API videos from scratch (POST /v2/videos) or from published templates (POST /v2/videos/fromTemplate).
• For personalized batches where variables drive names, roles, product details, CTAs, languages, or avatar choices.
• For QA related to lip sync, script pacing, pronunciation, avatar framing, identity disclosure, localization, download custody, and policy compliance.

What it can touch

• API endpoints: POST /v2/videos, GET /v2/videos/{video_id}, POST /v2/videos/fromTemplate.
• Inputs: avatar, avatarSettings, background, scriptText, scriptAudio, input array, templateId, templateData, brandKitId.
• Outputs: video id, status values (complete, deleted, error, in_progress, rejected, approved), download links, final MP4, and custody metadata.

Caveats

• Verify access levels: API access documented as available for Creator plans or above; test requests limited to 30 per day.
• Ensure consent for Personal Avatars and voice usage; policies prohibit impersonation or misrepresentation without documented consent.
• Avatar/voice sharing is Enterprise-only; deletion of an avatar does not remove it from existing videos.
• Private download links are time-limited; store final artifacts in the project manifest.
• Stay within policy limits for allowed uses (e.g., no political persuasion or deceptive impersonation).