01 · Engine

Load a pinned model configuration

Not loaded

Auto uses the conservative TTSGenerator operating policy, not an official hardware minimum. Official ST2/ST3 use verified Worker-managed model bytes; model initialization starts only after this button is pressed.

Model cache

Application-managed model assets

Not checked

Core and Voice components are tracked separately. A cached core does not imply that the selected Voice is cached.

Download Waiting for cache inspection
Verify / Cache Waiting for model assets
Initialize 0 sessions completed

No model request has been made.

02 · Synthesis

Generate speech locally

Detecting the best local model…

47 characters
Selected Voice Voice M1 · Not loaded

Diagnostics

Runtime evidence

Router decision

Phase A.3-5 timing, cache and lifecycle

Structured performance data
No performance run recorded.
Application cache Not checked

Environment

Raw event log

Waiting for an explicit model load.

Confirm model cache deletion

No matching namespace.