#!/bin/sh set -eu # Local transcription models download lazily on first setup; no external # native speech runtime or Python environment is required. exit 0