There are some pretty neat devices that are similar to audio recording devices that you can speak into and it will speak the translation for you. A friend of mine got one for a trip to Japan and apparently it worked pretty well!
100%, and are freaking fast...and pretty efficient, especially for high throughput applications tracking tons of changing data. We've found bitmaps handle updates, inserts, and schema changes really elegantly - and quickly (like sub-second). Depending on the use case, looking at cohorts MoM or YoY that freshness doesn't entirely matter, but computing cohorts in real-time hour-over-hour, or even minute-over-minute, for in-app actions/personalization is insanely valuable to keep users engaged (or what ever entity type you are analyzing and taking action on, whether it's devices, nodes across a network, etc).
Yeah in this type of architecture there is some sort of join required. Interestingly, @maycotte talking about the Molecula feature store – it serves real-time and historical features in the same storage layer. Significant performance gains with that approach.
We ran into this same problem trying to build and run a data management platform on it... IDK if you want to try what we built, but it's a hell of a lot faster and less cumbersome. We went from 30 ES servers down to 4 Pilosa servers.