I found the sine chords one very, very interesting. Truth is, those sounds should compress rather well, and be left intact. But they aren’t. Listen to the ghost file.
Now, here’s the cool part.
Go back and listen to the compressed one again. You will feel some bits didn’t happen. And they didn’t!
Hearing the ghost file tunes your perception a bit, somewhat altering your expectations, which renders the compressed version of the sound and it’s artfiacts more noticeable.
This guy gets at the lost sounds a few different ways.
Check out example 9! It appears a or may be the formant of Suzanne Vega’s voice is frequently REMOVED! (I have to go listen to her uncompressed some to be sure.) For many vocalists, this would be unacceptable, and frankly I’ve written here many, many times about how I don’t like vocals when compressed very much.
BTW, on a side note, it is possible to vary these in most people. Unless one has innate talent, or is made aware of this aspect of voice resonance, it’s a primal, basic thing that just happens. Great vocalists either have control of it, and manage that, or their natural state when singing just ends up consistent, and on that basis appealing.
A clear, formant, consistent, is one primary attribute by which we find we can differentiate vocalists regardless of language or style. Vega is notable for her very light use of it, instead singing largely with overtones and noise…
And she was the benchmark for the early mp3 compression code too. Ugh…
(nothing negative intended, nor implied. I happen to love Vega’s vocals)