Interesting question! Let's test:

JS Benchmark for several floor function alternatives

Results: Math.floor is about the same speed as other numeric methods for flooring a number, and all are far faster than you'll ever have to worry about.

Conclusion: Use Math.floor for readability, but these kinds of micro-optimisations really do not matter lol.

Is it faster to use '~~' than 'floor' or is it the same?

The description was updated some time ago to reflect this.

Random test are failing, and it is impossible to find issue, because generated input is not provided.

expected Array(16) to deeply equal Array(16)

Would be nice to provide input that wass passed to this test.

Not sure from where you're getting:

['Db'] -> -4 steps -> ['A#'],

...because if you're referring to the example in the description, that one is:

['Db'] -> -4 steps -> ['A']

and that is correct. A natural is indeed four half-steps below D flat.

You would achieve G flat by descending four scale steps below D flat (in the key of Db major), but the kata is not dealing with scale steps; it's chromatic differences (half-steps) that are being used here.

Since you are posting as an issue, I would ask that you clarify exactly what the problem with the kata is?

As a musician, it's strange to use both flat and sharp in one song. Additionally, you may have to make it clear that why ['Db'] -> -4 steps -> ['A#'], for it should be 'Gb' instead. Btw, it's a great Kata and I'm looking forward to see more music-relative practices here.

Thanks bro

It's amazing bro

To anyone who's confused by the description: the real difficulty in this kata is to observe how the process work.
It might have to take some time, but not difficult.
Just remember, every signal process is not necessary the same.

Good luck warriors!

nice!