Sleep as Android may keep the last few seconds (configurable) of noise detected in memory (buffer). When the recording threshold is reached, it can save the noise data in the buffer. Noises before the loud one that triggered the recording will be recorded.
This will be useful in providing context on the loud noise recorded. For example, if my kid open my door, asked if I was awake, then slammed the door after, I will know who slammed the door even though his voice is not loud enough to trigger a recording.
And more seriously: sleep apnea. A gasp triggers recording, but it’s missed, and all you have is a 10-40s clip of silence. Mintue after minute.
This night, after a physically very exhausting day, I had a worrying string of empty, silent recordings, timestamped 0:45, 0:46, 0:48, 0:50, 0:51, 0:52, 0:52, 0:53, …all the way to 1:41. I counted them, 48 separate short clips of silence. With a heart rate going from a steady 60 to jumping between 75-100. It’s slam dunk sleep apnea, the clues are obvious, but I don’t have a recording that I could show to my doctor. Only the silences in between the episodes. It’s kinda funny.
Time to dig up an old phone, set it to continuous sound recording, and visit a doctor ASAP.
I can totally relate as apnea sufferer myself. Sleep as Android helped me realized I have sleep apnea. From the noises, I learned I move every 10-30 minutes or so.
I hope you can solve your apnea problem. I tried mouthpieces, but got inconsistent results. What worked best for me is CPAP. Apnea is a terrible terrible disease and most sufferers does not even know they have it.