Fantastic question and data. Without committing to a clear answer now, may I engage you a bit of further discussion? I am an experimental physicist, just retired from 40 years as physics Professor at University of California, Irvine.
There are a couple of initial questions. 1.) How was the crankshaft suspended (attachment points and means). 2.) How was it struck: location, striking tool, and means of striking.
There are longitudinal, transverse, and torsional normal modes which may be of interest.
For the graph shown, it looks like you have a fundamental mode at 375 and second and third harmonics showing. The harmonics of a fundamental mode typically are somewhat close to integer multiples of the fundamental but maybe not exact integer mutliples.
The mobile phone microphone will have a frequency dependence on its detection capability. Typically most humans do not hear much above about 12,000 Hz so phone microphones don;t need to be sensitive above that. A human scream, indicating danger, will have lots of components in the 2,500 to 4,000 Hz range, below that range does not cause human danger responses as much. Below about 100 Hz are difficulty to measure with small microphones, typically below about 200 Hz will be under-measured by small microphones unless compensation in analysis is used. For example, people don’t hear 50 Hz signals well in the ears but can feel them in the center of the body as a thumping.
I have a loose Mark V 3 1/2 crankshaft and can make comparison measurements.
I will check in some engine design books to see about crankshaft fundamental and harmonic frequencies. I suspect this question is well understood within the community that designs crankshafts.
Thanks for posting this topic!