If I'm deaf and I pay for an app to tell me when an alarm is sounding, I have a certain expectation that it has been rigorously tested and will work without fail. And I'm a software engineer who understands the limitations and failure modes of such applications on non-real-time, commodity kit.
Please be extra careful about how you market this. I absolutely applaud your ingenuity, but I'd feel differently if I found out my elderly, hard of hearing gran was relying on any app on her smartphone to wake her if she left the stove on and went to bed.
I expect a proportion of your customers may be less technical users (or well-meaning relatives) with little understanding of the limitations and risks.
However, it is indeed an interesting idea. There may already be something on the market, but what about developing it into a physical device running on dedicated hardware? Crowd-fund an "alerter badge" or pin or whatever, and budget heavily for testing & verification. You could be onto a winner.
EDIT: Please don't be put off altogether by the comments. It's simply that non-real-time, non-failsafe smartphones aren't built for life-critical applications like this. The idea isn't a bad one - you're just targeting the wrong platform.