Two Sources for Training the AI: Memory and Material

Source 1 — the family's memory

The biography, stories, life feed, and voice transcripts automatically become the copy's knowledge base. The fuller the page, the more faithful the conversation.

Source 2 — material

Separately, you upload voice audio for cloning, message histories for the speaking style, and face photos for the 3D head.

Together — a whole image

The two sources complement each other: memory provides facts and character, the material provides voice and face. That's how a recognizable copy comes together.

Where to start

First fill out the memory page — it's the first source, and the most important one. Add the voice and face material as you find clean recordings and good photos. There's no need to rush: the copy can be trained further over time.

• Source 1 — the memory gathered on the page.
• Source 2 — voice, message histories, face photos.
• Memory gives character, material gives voice and appearance.
• Can be trained further gradually.