Putting your efforts and money into 20-20KHz +/- 0dB is counterproductive for several reasons.
1. Flat frequency response is overrated in the sense that it is the distortion that we listen through the most easily. Think about a live concert. Between concert room interactions to the sound guy doing EQ boosts and dips, live music is never flat, yet it is the gold standard for fidelity because it is real. We are more used to listening past frequency nonlinearities than any other type of distortion.
2. Flat frequency response would be nice if it were achieved from one end of the signal chain to the very end, but it never is. So having a flat response for playback only preserves the last tweaks during mastering. Some recordings are mastered to sound good on an iPod or a car radio. Is this how you want to hear them through a 20-20KHz flat system? Personally, I can't BEAR to hear Motown CD reissues on a hi-rez system.
3. Some means of achieving flat frequency response wreck other sonic cues for which human hearing is more sensitive. For example, heavy-handed equalizing messes up phase relationships, which we use to identify timbre and location. That's one reason 1st generation CD players sounded so bad--they used a "brick wall" analog filter to shut down any frequencies above 20KHz, and wrecked the phase relationships of the audible overtones from 5-20Khz in the process.
4. 20KHz is nowhere near high enough to achieve realistic-sounding fidelity. It's not that we can easily identify test tones above 20KHz, but the upper frequency response also defines the rise time of every reproduced note. If my math is correct, a 20Khz top end limits rise time to 1/40,000 of a second. Listen to a system with electronics rated out to 150KHz and the music sounds more lively with faster and more dramatic transients. Good transparency requires speed. Look at how the Gram Slee Era Gold is such a highly regarded phono stage for the money. The designer is adamant that having a frequency response out into the MHz range is essential for preserving speed and upper octave overtones and therefore, timbre.
1. Flat frequency response is overrated in the sense that it is the distortion that we listen through the most easily. Think about a live concert. Between concert room interactions to the sound guy doing EQ boosts and dips, live music is never flat, yet it is the gold standard for fidelity because it is real. We are more used to listening past frequency nonlinearities than any other type of distortion.
2. Flat frequency response would be nice if it were achieved from one end of the signal chain to the very end, but it never is. So having a flat response for playback only preserves the last tweaks during mastering. Some recordings are mastered to sound good on an iPod or a car radio. Is this how you want to hear them through a 20-20KHz flat system? Personally, I can't BEAR to hear Motown CD reissues on a hi-rez system.
3. Some means of achieving flat frequency response wreck other sonic cues for which human hearing is more sensitive. For example, heavy-handed equalizing messes up phase relationships, which we use to identify timbre and location. That's one reason 1st generation CD players sounded so bad--they used a "brick wall" analog filter to shut down any frequencies above 20KHz, and wrecked the phase relationships of the audible overtones from 5-20Khz in the process.
4. 20KHz is nowhere near high enough to achieve realistic-sounding fidelity. It's not that we can easily identify test tones above 20KHz, but the upper frequency response also defines the rise time of every reproduced note. If my math is correct, a 20Khz top end limits rise time to 1/40,000 of a second. Listen to a system with electronics rated out to 150KHz and the music sounds more lively with faster and more dramatic transients. Good transparency requires speed. Look at how the Gram Slee Era Gold is such a highly regarded phono stage for the money. The designer is adamant that having a frequency response out into the MHz range is essential for preserving speed and upper octave overtones and therefore, timbre.