Okay, can someone really explain this to me. I've had multiple conversations with dannys1 on here and others and I've yet to understand this flat speaker lingo. Some say it's not flat. Others say it's equalized. Apple claims it's balanced, I think. Are all these things the same or different things? Is flat/neutral/EQ/balanced all the same?
I'll try.
There are two kinds of "flat" - a measured frequency response test result that produces as close as possible to a flat line on a graph, and the subjective perception that the sound is "uncolored." Well, there's a third kind of flat; uninflected/boring. That's not what we're talking about here.
All speaker systems are "equalized" - this is a function of compensating for the native frequency response of each component part, the design of the crossover system (equalizer that separates frequency bands for sub-woofer, woofer, midrange, and tweeter), the contribution of the cabinet design on frequency response (both on-axis and off-axis)...
A key goal in speaker design is to keep the speaker system within basic norms (relatively flat frequency response graph), while maintaining an identifiable brand, so that regardless of what speakers were used when producing the sound, it'll still sound reasonably good when played back in the home.
"Flat" is nominally a frequency response curve that looks like a flat line, except at the upper and lower extremes of response, where there's an inevitable fall-off in response. Flat, however, is still relative. It's fairly simple to produce flat response in electronics, such as +/- 0.1 Db from 20 Hz to 20,000 Hz. Presuming that transient response is also sufficient (again, fairly easy to do with electronics), that would produce essentially uncolored sound. However, speakers, by their nature, are held to a less stringent standard - say, +/- 3 Db from 20-20K, to allow for dips at crossover frequencies, spikes/resonances due to physical characteristics of cabinet, speaker cones, voice coils, etc., and fall-off at the more extreme frequencies. It's a lot harder to produce "flat" in speakers and microphones than it is in electronics.
Engineers and producers like to have a consistent point of reference when moving from studio to studio. They often bring their compact, "flat" reference speakers with them, to compare to the studio's main monitors. When an audio pro talks about a "flat" reference speaker it often means that the speaker lacks noticeable coloration, rather than describing the results of a frequency response test - two speakers with apparently identical response curves can still sound different (which is why subjective listening tests are part of speaker evaluations). This, of course, is a subjective judgement. One engineer's "flat" may not agree with another's.
I'd agree that HomePod has a colored sound. I find the coloration pleasing, another may not. Part of the coloration comes from the sound being atypical - highly processed. Since it's an atypical design, that's not a surprise. Unless and until HomePods and speakers of like design are widely deployed in homes, it's not appropriate to use them as a general-purpose reference.
Overall, the bass response is what's most remarkable. I'd like a more transparent-sounding high end, but still, it sounds remarkably good for both classical and pop, which is not always easy to do. Violas and cellos in chamber music recordings sound fabulous, rich and well-defined, very close to "live," which is often not the case. This may seem to be a comment on mid and bass response rather than highs, but highs are essential to provide proper definition.
I used HomePod last night when watching Wonder Woman on my 12.9" iPad Pro (AirPlay was very easy to setup). That was a mind-blowing experience. The bottom end was comparable (though obviously not identical) to movie theater sound, and for the first time, I was disappointed with the sound coming from that iPad.
My only complaint has to do with Hollywood sound mixing practice - they seem to add a fair amount of "bottom" to female voices, which coupled with all the room-shaking low-end sound effects of an action film can make those voices harder to pick out. I encountered the same problem on HomePod as I have in theaters. On the much thinner bottom end of the iPad, voice intelligibility wasn't an issue.
My instinct in the studio would be to reduce the bottom end of voices so that the dialog is more intelligible, but nobody in Hollywood is paying me to mix their films. ;-) So, for pure voice intelligibility, iPad wins. For "truth" to the original mix (for good or bad), HomePod wins.