What it does
This runner explored a repeatable way to request short speech samples from a local model. A small CPU sample was verified during the experiment; dependable fresh generation and acceleration were not established. It is not the current creator narration pipeline.