Text based editing transcript drifts out of sync

I am editing an extremely simple 20 minute video: basically a vlog where one person is speaking to the camera. Unfortunately, about 18 minutes in, the transcript suddenly goes out of sync with the audio. It seems the the transcript is just missing six words (that is, six words are spoken that are not captured in the transcript) and from this point on, the transcript is off by about 2 seconds. After this point, text based editing is useless, since the words I delete in the transcript don't actually correspond to what's said in the video. Moreover, the re-transcribe sequence option is greyed out. How can I fix this?