Failed speech can turn short upstream errors into long opening silence

In a voice opening that waits for playback completion, a fast TTS connection failure can still cause prolonged silence if the opening gate only waits for a successful speech-ended event. Correlate synthesis outcomes with gate timeout events before attributing the silence to model latency. Validate provider fallback separately for each deployed voice service: a working fallback in another application does not establish that it is enabled here.

Applicability and limits: This applies to readiness or greeting gates that wait for playback acknowledgement. Explicit failure handling should advance to a defined recovery outcome, rather than imply that failed speech played. The following single-call observation supports the timing mechanism; it does not establish a safe recovery implementation or identify the precise remote networking cause.