Hello, Nine8Six.
Good idea and it will work. I say that having worked as a senior development engineer for hardware and software in various digital products, including a lot of audio. I'm a 25-year+ member of IEEE and AES.
You may be underestimating the amount of code that has to be written, not to mention the extensive debugging required to make it reliable. In this case, the amount of code will be driven by the large number of digital filters needed and the interaction among them. This is because motors are noisy places. Picking out the relevant signal from the noise could only be achieved by "self learning" algorithms "hearing" more than a few failures. Lacking any database of failure sounds, the software will have to determine the onset of failure based on changes from "normal." Of course "normal" will vary from motor to motor and across a large operating range.
In short, this is a hard problem that does not lend itself to generalization. Some strategies may be transferable to other motors in other models of cars, but most of the code will not be reusable.
If you were the Director of Engineering at Porsche and asked me as a consultant to bid for this software work, I would estimate 2 man-years. At my rates, that is over $300K.
Good luck with it.
Cheers,
Dave
Earth Tour
Helios 1
Threaded Mode
