Suppose I can record audio of a vehicle near different points (e.g. beside the engine or exhaust) during different activities (e.g. revving, idling, starting), and suppose I am well-equipped with mathematical (e.g. Fourier analysis) and computer programming knowledge (e.g. Python).

What can I learn about the health of a vehicle from studying these audio files (e.g. spectrograms, mode decomposition, tomography, unsupervised learning, etc)?

Motivating Examples:

1 Answer 1


There are many defects that, when advanced enough, are clearly hearable.

They cover not only those areas where moving parts are involved (suspension, drivetrain and engine), but also defects in body are often hearable (while driving). Odd sounds are clear hints that something is wrong. The experience to interpret those sounds is neither trivial nor easily documented.

An detailed list, inclusive every root cause and sound description might be a big task. In my understanding this question is similar to those where one asks for an complete list of those secret manufacturer fault codes, so he can produce (and sell) his own obd app, without paying for the entire work (collecting the data).


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .