Thanks @mrueschman and @DanMobley. It seems like maybe 20% of sleep labs record simultaneous audio. I'd be up for working directly with a lab if you know of any who record audio and participate in research.
Looking at the montages of the available data sets 2 include a snore channel, but it's sampled 100-300 Hz. Low Hz like that isn't actually audio, more just an indication of the existence of sound. I'm working on a project that analyzes audio to detect apneas and hypopneas. I don't know the exact Hz I need for my analysis, but I suspect 300 Hz is too low.
Have you seen any datasets with a corresponding audio signal of significantly higher Hz?