Monotonic word timestamps are a format check
Monotonic word timestamps are a format check, not independent evidence that words occurred at those times. Forced-alignment libraries can repair nonmonotonic model outputs into plausible ordered intervals; validating only the repaired output can hide the very failure being checked. This applies when using alignment to resume interrupted synthetic speech. Preserve raw timing diagnostics when available, reject out-of-audio or collapsed spans, and validate claimed boundaries against independent acoustic evidence. Keep alignment estimates separate from transport playback receipts; neither proves listener comprehension.