AI Identity Builder

🎙️

Vocal Profile Auto-Cloner

Upload raw sample audio to automatically spin up a brand new ElevenLabs voice model array.

Doctor / Target Profile Name

Vocal Sample Dataset (.mp3 / .wav)

📤 Click to select raw 30s - 2min uncut audio sample

🎬

Link existing HeyGen studio tokens and profile codes directly into the local Postgres database.

Link to Existing Voice Target

Production Avatar ID (From HeyGen)

📸

Train a custom, drift-free Flux visual identity to safely bypass สสจ medical presentation rules.

Unique Prompt Trigger Token Phrase

Reference Image Folder Dataset (.zip)

📦 Click to select ZIP archive containing 30-50 high-res photos

⚠️ Mandatory Dataset Schema (Anti-Drift Rules):

ZIP File Only: Compress all photos directly into a single package.
Frame Mix Ratio: Include roughly 50% Close-ups (Face), 30% Mid-shots (Waist-up), and 20% Full-body shots to completely lock in natural human proportions and prevent body distortion.
Angle Profiles: Ensure a clean multi-dimensional split of Center shots, 45-degree Left profiles, and 45-degree Right profiles.
Style Variation: Use a dynamic mix of different clothing outfits, lighting setups, and background locations. Avoid shooting everything against a single plain studio wall!

📡

Operational tracking index of custom weights deployed across the generation pipelines.

⏳ Querying active weights from database...