You some controls: for example, the same mass/second of water - your water pressure might differ between hot and cold sources.

Second sample has significant hi-hat style sustain, since I know it is hot water it should be a steam. First sample has really little amount of this sound which might be produced from bubbles. First sample is really interesting because it has 2 bass lines: first starts at 900Hz and rising to 100Hz, second starts at 600Hz and quickly dropping to 400Hz and keep lowering slowly to 300Hz. Definitely not harmonicas, it is something like 2 independent resonating frequencies for something inside of glass-and-water sound source. Bassline of second sound is singular fat line; firstly it starts at 700Hz and slowly grows upto 1kHz.

Probably the 2 basslines from the first samples (cold) was something that bothered the dog, maybe the thing is in rich sound such as chord vs single note. I don't care about bubbles but when I will have an access to a nice microphone I will try to record more samples of this. The hard part is how to remove water from the glass between 2 or more experiments because it is important to have really similar setups without moving the glass even slightly.