Comment by oflannabhra | Hacker News Reader

Comment by oflannabhra | Hacker News Reader

I'm no audiophile, and I don't plan on getting a HomePod, but the interesting tidbits to me were the following:

>Apple uses Balanced Mode Radiators (BMRs) instead of industry typical tweeters. They have a response range of ~250Hz-20kHz, whereas typical tweeters have a range of 2kHz-20kHz. Here is gif of a BMR compared to other speaker technologies [1]

>Apple applies Equal-loundness contours[2] to equalize absolute energies of loudness to perceived loudness by the human ear. That is, the dB of sounds in the 2KHz-5KHz is decreased by several decibels, because the human ear is more sensitive to them.

>They recommend putting the HomePod on a small stand (5 in), because even the room correction processing Apple is using is unable to compensate for echoes that originate so close.

>There is apparently some agreement within this community, at least, that if Apple made a HomePod Plus with a larger subwoofer to allow reproducing sounds down to 18Hz, they would essentially beat the entire high-end audio market.

[1] - https://gfycat.com/BiodegradableNiftyKoala

[2] - https://en.wikipedia.org/wiki/Equal-loudness_contour

> That is, the dB of sounds in the 2KHz-5KHz is decreased by several decibels, because the human ear is more sensitive to them.

I would expect recorded material to already account for this. Does anyone know why Apple finds it needed to further apply sound shaping to recorded sound?

(That is, I would expect audiophile-grade equipment to best mimic the monitors on which most recordings are mixed, which presumably is a flat loudness curve.)

>(That is, I would expect audiophile-grade equipment to best mimic the monitors on which most recordings are mixed, which presumably is a flat loudness curve.)

Not necessarily, because that assumes the audio was mixed correctly to begin with. The audio will have to be played back over digital and analog broadcasts with varying dynamic range characteristics, not to mention various output technologies, including $5 headphones. And not to mention how the shape of each persons ear affects the response curve in different ways.

There is no way for an engineer to produce a rendition that sounds great in all of those scenarios, making 'mimicking the monitor' a pointless pursuit to begin with - even IF you were an audiophile listening to 400MB DSD files.

>Does anyone know why Apple finds it needed to further apply sound shaping to recorded sound?

I'd say the vast vast majority of people want more to have the audiophile label/association than actually listen to music with a flat curve, because most music will sound like crap.

Sorry for the rant :P

The perception of equal loudness changes with both sound pressure level (SPL) and frequency. Are you playing the music louder or quieter than the SPL it was mixed at? Then the balance changes.

Thank you, that makes a lot of sense.

Because you are changing the volume. If you read about it further, it's adjusted on the basis of how loud you turn it up.

Long and short of it is that if you turn the volume up, and your want it to sound the "same" but louder than before, you need to turn different frequencies up more than others, to account for the human ear's varying sensitivity to different frequencies.

You'd think so, but recordings are usually mixed to sound good out in the wide world. The most famous mixing monitors are famous for sounding awful.(https://www.soundonsound.com/reviews/yamaha-ns10-story) The thinking there is if you can make the song sound good on those monitors, it'll sound good anywhere.

A lot of songs are mixed with less bass than ideal, because the engineer knows that everyone and their grandma has that bass boost button permanently down on their stereo. Or people are using beats headphones that crank the low end.

The ideal mix also changes over time. Part of that is fashion. Every period has its own ideal sound. But another part is the equipment the average listener has.

Ever have a beater car with a terrible stereo? Music from the 60s will still sound great because it was mixed to sound good on transistor radios. You can hear the bass guitar even though there's no low end on your setup. Anyone who had a K car in the 90s has a love for CCR. That's a fact.

In previous decades music was mixed to sound good on wood encased speakers, which have their own resonance characteristics. For modern music, you assume plastic. You also have to assume that a large portion of listeners are going to be using apple earbuds, or some other cheap earbud.

When working on a mix, you usually have a few different speakers to choose from in the studio. But the final test is always how it sounds in a car.

I agree with everything you said, but were Apple compensating for that, I'd expect them to use a curve weighted to crappy studio monitors/Beats headphones/etc.

A couple other commenters noted that perception-weighted curves vary with total SPL; louder sounds sound different. I had never thought about that before but that makes sense to me, that Apple would apply a dynamic adjustment that studios can't.

Oh, that's an excellent point too! Bass boost and loudness buttons on stereos are meant to give you a way to make music sound the same at a lower volume, but most people just set it and forget it.

With fully digital players, I guess there's no reason to not have dynamic EQs that adjust along with the volume. I'd think you'd see this more often, but I guess even basic EQs aren't available most of the time, so it's just not a feature that most people would care about.

Reply on news.ycombinator.com