TTS Canary Runner
Capabilities
- run_canary: execute small-sample TTS runs per voice/model.
- evaluate_quality: score latency/artefacts/volume and compare to thresholds.
- trigger_rollback: stop rollout and flag failures if thresholds exceeded.
Dependencies
- reliability-budget
- ops-chief-of-staff (optional scheduling)
Inputs
- voice configs, test scripts, thresholds.
Outputs
- canary report with pass/fail and metrics.