RE-Avatar
Research/00-index.md
RE Avatar — Research Index
Running log of all knowledge sources consumed while building RE. Each source gets its own note in this folder. This file is the index.
HeyGen API Sources
| File | Source | Status |
|---|---|---|
| (pending) | HeyGen Interactive Avatar API docs | — |
YouTube / Video Sources
| File | URL | Topic | Status |
|---|---|---|---|
| (pending first URL from Arijit) | — | — | — |
Key Findings (running summary)
(Will be updated as sources are processed)
HeyGen Interactive Avatar — What We Know So Far
- Platform: HeyGen — AI video generation + interactive avatars
- API version: v2 (streaming, WebRTC-based)
- Session lifecycle:
create session → start → send task → stop - Core SDK:
@heygen/streaming-avatar(JavaScript/TypeScript) - Python: REST API calls only — WebRTC rendering is browser-side
- Key inputs to avatar session: avatar_id, voice_id, knowledge_base_id
- Supports custom knowledge bases (uploadable docs)
- Supports STT modes: server-side (HeyGen) or client-side (you handle audio)
Sources Queue
(Arijit drops URLs/videos here — I pull, transcribe, extract, and file)
How to add a source:
1. Drop URL in chat
2. I create Research/NN-<slug>.md with extracted notes
3. I update the index table above
4. Key findings roll up to the running summary