I have always argued that the voice is the sound we recognise the best because we are most attuned to it from our mother’s voice as a baby and onwards through life to wanting to hear the exact characteristics of a favourite singer’s voice.
Consequently, I have always been beyond fussy about voice and the midrange. E.g. Croft amps, classic BBC speakers and, of course the Quad ESL 57s.
Whether a bass guitar should sound tighter (very venue dependent) or a synthesiser more focussed in the original recording is neither here nor there if voices do not sound simply superb IMO.
For this reason, my speakers over the years have had to excel in this area. The V or U shaped frequency response curve seems absurd to me especially if you listen to vocal music…
Also, the prominent bass it requires will sound slower with more overhang than was intended. I prefer a fast, leading edge to the bass notes without the bloom.
And yet, I have been a little surprised lately by how some recent speaker designs in shows have seemingly clearer, better defined midrange instruments than my all-time favourite speakers while at the same time having inferior, more synthetic-sounding vocals!
Then it dawned on me, when I was studying some speaker crossovers, that often the tweeter is tipped-up to give ‘better definition’. One demonstrator pair had 3 ohms less resistance on the tweeter than the original crossover diagram to impress the gullible. This ‘improves’ midrange definition and attack but tends to make voices sound less real and more robotic! (Never mind the fact that many ears prefer a dip around 2 KHz over the flat response…)
A manufacturer’s trick which relies on impressiveness over naturalness in vocal tone.