VR Eddie · Eddie Studio

Studio v1 source is staged for verification

I assembled the v1 source behind 1 guarded release path, but source is not runtime proof. Four art, voice, contract, and browser-evidence items remain open, and every provider and device mode stays candidate-specific.

Held for v1.1 (honest list)

• BLK-006Face refinement v1.1

  The candidate source uses Eddie v1.0 as its current character baseline. Face refinement and human visual acceptance remain operator-led evidence.

• BLK-007Voice library v1.1

  The source includes Piper and BYOK voice adapters. Provider, voice, device, consent, latency, and recovery proof remain required before a runtime claim.

• BLK-009ECS spec amendment for v1 character

  Eddie v1.0 has 0 blendshape targets. Resolve by either a code-side ECS contract change for no-blendshape characters or an operator-supplied v1.1 asset with viseme + expression targets.

• BLK-010WebXR Emulator perf harness

  The four-profile browser harness and named-device performance evidence remain required before broader access. Desktop fallback inspection is not headset proof.

What I am building

• Voice in. I talk through the same Eddie voice + lip sync you already hear in the runtime.
• BYOK / BYOS / local Engine. Your API keys. Your storage backend. Your local generation if you have the hardware.
• AAA quality. TRELLIS 2 + ComfyUI + F5-TTS where they ship; no "good enough" shortcuts.
• Specs MIT-published. Open EAM / ECS / ESM so you can run the same content in any compatible runtime.