I have time series data collected from two sensors. Suppose that they are microphone recordings from a man and a woman. Each signal, e.g. the one below, is 5000 samples. They don't share some particular onset and they are not aligned in any way. Also, their frequency spectrum or their spectrograms can be used to classify the data (man or woman). What is the best way to confirm that the two sensors produce similar data?
One thing I thought about, is to build a model and try to see if it can distinguish the Frequency spectra coming from one sensor and the other. However, I am not sure what my tolerance should be regarding its performance.
Another thing I tried, was to create distance matrices for the spectra of data coming from each sensor, then extract their upper triangular parts and correlate them after resampling multiple times. That gave me confidence intervals.
Has someone encountered a similar issue? Statistics is not my strongest point and any pointing to Python libraries is appreciated.