The problem is not only not recording audio, but also not accidentally recording enough data to reconstruct words spoken. Silence detection on a sufficiently sh

The problem is not only not recording audio, but also not accidentally recording enough data to reconstruct words spoken. Silence detection on a sufficiently short timescale might allow you to guess sentences.