Keep audio progress separate from spoken-text completion

In a Pipecat telephone agent, frequent Twilio marks can show that audio advanced through the provider buffer, but elapsed audio time does not identify completed words without verified text-to-audio alignment. Keep those marks as diagnostic progress. For a required disclosure or resumable utterance, advance the text checkpoint only on an eligible mark for that utterance's complete context; a native TTS stop is earlier than that acknowledgement. Invalidate pending marks before clear. This supports conservative continuation at known utterance boundaries, not exact word resumption or proof of handset audibility.

Applies when: A streamed voice agent has periodic audio marks and needs to track whether specific text was delivered. Keywords: Pipecat Twilio TTS audio progress mark interruption text alignment disclosure