To be fair, if I had to guess, I think the measurement of interest is for American translation of said app already in development (or developed) by some Chinese devs. If anything I'm kinda impressed at the production quality; can't seem to find anything about the developers though.
As a person with some experience with mocap data analysis, the ad copy is mildly furiating (http://getmaoapp.cn/en/how-it-works/). But I suppose they do what they gotta do to impress the clientele.
I swear my cat gets frustrated sometimes when she's repeatedly trying to tell me something and I keep guessing the wrong thing (or worse, ignoring her) - so much so that she sometimes hits me... However, I believe that any human will probably effortlessly outperform a mobile app at cat translation for the foreseeable future.
Maybe, maybe not. A computer could pick up subtle shifts in body positions or vocal tone and pace that a human can't. Like the work to visualize a heartbeat and other bodily functions, by magnifying deltas that are too small for the human eye to detect. http://people.csail.mit.edu/mrub/vidmag/ Something like that could certainly conceivably work for cat body language and voices.
It would indeed be interesting to see some info on systems trying to accomplish this. I would argue that the cat owner gets better data in the first place because they are physically present, while a mobile app has to reconstruct a lot from video input.
Then there is the problem of interpreting the intent of the cat itself. I would say there are a few different modes of communication with varying degrees of obviousness.
First, the app would probably be good to determine basic mood, such as fear or joy. But then again, any human could do that just as well. More nuanced communications of mood and intent can vary from cat to cat, giving the owner-human the edge here as well. I don't need an app to tell me that my cat feels 57% optimistic about her day, 32% playful, and wants more social contact with a 75% probability. As a family member, I can also use other heuristics that an app can't, for example I can factor in what happened yesterday, or other specific data that is hard to generalize programmatically.
The other category would be purposeful coordination with the human. In this, the cat explicitly addresses a human with the intent to communicate. This is probably where the most misunderstandings happen, and hence the best opportunity for the app to shine. As a lifelong "cat observer" I'm pretty good at reading a lot of the cues instinctively, but these too vary a lot from cat to cat. Usually the cat also tries to show what the problem is. For example, she literally walks me over to the door, or to the place where food is stored, or to where a toy is kept, or to a dirty litter box if I forgot to clean it. My cat also regularly points out other things that are clearly my fault, such as rain or the wrong food flavor. Once again, this is not hard to recognize.
In general I think this is a fascinating area, not because it's about cats, but because of the implications of communicating with minds that are not human.
Although the style of this website is sufficiently 'Western' that these people are probably trustworthy, I would recommend some caution when it comes to signing up for their mailing list. My experience is that Chinese companies—even the ones you might think are respectable—don't know the meaning of the word 'unsubscribe'.
Step 2: Post on hacker news