let $\Theta^+$ be the pseudo-inverse of $\Theta$.
Recall, that if a vector $\boldsymbol v \in R(\Theta)$ (ie in the row space) then $\boldsymbol v = \Theta^+\Theta\boldsymbol v$. That is, so long as we select a vector that is in the rowspace of $\Theta$ then we can reconstruct it with full fidelity using the pseudo inverse. Thus, ...
The Raspberry Pi ZeroW with V2 camera doesn't quite have the specs you need, but it is close: https://picamera.readthedocs.io/en/latest/fov.html#sensor-modes
Here's a waterproof enclosure for it: https://www.youtube.com/watch?v=5CWksss_5lQ
random forest's feature importances are not reliable and you should probably avoid them. Instead you can use permutation_importance: https://scikit-learn.org/stable/auto_examples/inspection/plot_permutation_importance.html#sphx-glr-auto-examples-inspection-plot-permutation-importance-py
The task of isolating 2 or more speakers is called speaker diarization,
here a list of softwares and useful resources.
Once you have the 2 or more audio files containing the individual voices, you could run some speech-to-text network that also outputs time information.
It's difficult to say without knowing what your data looks like but from the numbers it seems too less and the images might be too similar to one another or very different. In any case, I'd have checked using other networks like Inception and decreasing learning rate even further (say 0.0001) to not mess with the Imagenet weights if your data is not very ...
Neural networks can have a lot of different structures. CNNs can have a number of parameters that ranges from a few thousands to several millions.
In general you aim to increase the number of filters and reduce the first 2 dimensions, as you go deeper in the network.
So if you had Conv -> pool -> Conv -> pool -> ... , you could do for example ...
You are using Dense layers, try 1d convolution instead.
Have you tried a different activation function such as softmax and instead of Binary cross entropy try MSE loss? Are all your inputs between 0 and 1? Also, I think your noise amplitude is too much in 2nd and 3rd case as compared to the actual signal. Can you try training on different types of spectra ...