Apple Homepod in-room measurement

[RayDunzl]

Does anyone think that LF loudness compensation is too aggressive at lower volumes?

Not me.

As I see it, it only comes into play at higher volumes.

I don't see a low level boost (those traces are rather flat), I do see higher level progressive cut.

 

[Phelonious Ponk]

How the Homepod DSP separates the signal and sends it out to the multiple tweeters is kinda a black box situation for how it attempts to reproduce a REW sweep signal. I felt confident that I could at least get a decent loudness compensation measurement. Does anyone think that LF loudness compensation is too aggressive at lower volumes?

Some material seems to be a bit bass heavy at low volume.

 

[Cosmik]

Some material seems to be a bit bass heavy at low volume.

That's what I'd do if I were them - give a good impression of bass when it's easy.

 

[12B4A]

You're right Ray. The boost doesn't look as dramatic on the lower volumes of raw sweeps. I was overthinking it with the loudness compensation plot. I guess I'm just accustomed to flat at all levels and the milder Audyssey dynamic EQ when I feel like it. I just hope that with stereo pairing, the Apple audio engineers decided to carefully unbalance the bass at each pod to make the room sound less boomy. Still reserving judgement until then.

 

[Phelonious Ponk]

You're right Ray. The boost doesn't look as dramatic on the lower volumes of raw sweeps. I was overthinking it with the loudness compensation plot. I guess I'm just accustomed to flat at all levels and the milder Audyssey dynamic EQ when I feel like it. I just hope that with stereo pairing, the Apple audio engineers decided to carefully unbalance the bass at each pod to make the room sound less boomy. Still reserving judgement until then.

That would be the opposite of the Fletcher-Munson curve that was Apple’s model and objective. Something is very wrong here...

 

[Phelonious Ponk]

Measurements posted on reddit and Fast Company disagree with these.

 

[12B4A]

Fast company's method was using white noise and I didn't spot at what level they were measuring at. I have no idea if that's more or less correct but it's certainly different. Also in all these cases, it's an in room measurement of a device that changes its output based on what it senses from the room and at what volume level it is set at. I wouldn't publish data from any of these methods at AES.

 

[Phelonious Ponk]

Fast company's method was using white noise and I didn't spot at what level they were measuring at. I have no idea if that's more or less correct but it's certainly different. Also in all these cases, it's an in room measurement of a device that changes its output based on what it senses from the room and at what volume level it is set at. I wouldn't publish data from any of these methods at AES.

Is there a method for measuring a device that changes its output based on what it senses from the room and at what volume level it is set at? I’m following out of curiosity, or maybe just because I cant help myself, but it really doesn’t matter. I’m post-audiophile at this point, and this thing is much better than forward-facing boxes for the way I listen these days.

I’m listening through a pair of JBL LSR 305s and a 12” sealed SVS sub. No room treatment or digital correction. I’ve currently got the HomePod on my TV stand between the two of them. It may be difficult to measure, but it sounds really good. The sub, obviously, goes deeper, and sounds more controlled at high volume. The whole JBL/SVS system is obviously capable of greater volume. If there was a way to measure the pod that would give us a good comparison, we’d probably find out the JBL/SVS system is more accurate. If I had two pods running in stereo next to it, I might even like the sound of it better. But I guarantee you it doesn’t sound better 150 degrees off axis, in my dining room. And I listen from there, and the kitchen, and the chair in the corner of the living room, way off axis. I’ll buy a second HomePod when the Airplay update arrives. And if they sound as good as I think they will, they’ll get the stands, and the JBLs will go to the office, wired to my mixer. However they measure.

 

[Blumlein 88]

Is there a method for measuring a device that changes its output based on what it senses from the room and at what volume level it is set at?
snip

Yes. The way I know is tedious. It does work however. I'm sure there are probably better ways. Unfortunately I am short on time to respond.

In short play and record music at different volumes. Line up and match sections of the music. This need not be to the nearest bit, but close. Do an overall FFT for that passage. Adjust for the level difference in the mid band. Subtract the value for each bin in an FFT. 1024 FFt is probably enough for this. You'll see how response changes from on condition relative to another. You can dump the bin values into a spreadsheet and automate that part. So we aren't talking manually doing several hundred FFT bins.

 

[RayDunzl]

Measurements posted on reddit and Fast Company disagree with these.

Here's a link to one set of those measures, the Reddit guy:

Start here:

https://forums.appleinsider.com/dis...homepod-is-100-percent-an-audiophile-grade-sp

https://www.reddit.com/r/audiophile/comments/7wwtqy/apple_homepod_the_audiophile_perspective/ (File link in part 4, scroll down)

https://drive.google.com/file/d/1PcI_deY7tgDBhs7ra5S7m2gj3ezZmv8n/view

I think his measurements are at 5 feet. I think he is adjusting the signal level at REW and not using the Homepod Volume Control, based on the file names (100Vol). That appears to make the bass not compress like we see when the Homepod vo;ume control is used.

He used a UMIK-1 and REW to take the measurements.

Distortion on the file with about 85dB spl in-room, 100% Homepod Volume, -12dB input tone level.

Looks familiar to me.

I posted this simplification (THD only), similar test sweep SPL, using 124BA's measurement:

 

[RayDunzl]

Reddit guy SPL:

Since he seems to be varying the input tone level, and not the Homepod volume level, there's not the same bass compression noted, until the highest trace (-12dB input) gets compressed to the same level as the -16dB trace. A similar condition might be seen on the high frequencies, maybe it's rolloff at the top, but there's compression of the loudest (for sure) starting about 15khz bringing it close to the level of the next loudest by about 18kHz.

 

[RayDunzl]

Reddit Guy Step Response:

That's enough for tonight.

I think the Reddit Guy measurements validate the look I took at 12B4A's files.

 

[Soniclife]

Good to see some sort of consistency between different tests.

Regarding the bass distortion, in the white ring oil your wood articles there is the suggestion it is using what it rests on to generate bass, either deliberately or accidentally, which would almost certainly be highly distorted if true.

"The speaker requires a flat and solid surface in order to function correctly. Placing it on a coaster or mat negatively affects the audio quality, diminishing the bass response of the speaker – one of its key selling points."​

Measurements on different surfaces would be required I suppose to understand it. It may explain the step response if the surface it's on is excited.

 

[svart-hvitt]

Fast company's method was using white noise and I didn't spot at what level they were measuring at. I have no idea if that's more or less correct but it's certainly different. Also in all these cases, it's an in room measurement of a device that changes its output based on what it senses from the room and at what volume level it is set at. I wouldn't publish data from any of these methods at AES.

I suggested ASR members agree on measurement method, standardize measurement output and send these standardized methods over to @amirm for averaging and analysis.

Could be fun to read about such a meta test

PS: This could also be the one measurement exercise that puts ASR on the radar of a greater audience. Apple is big.

 

[Phelonious Ponk]

Well, I guess I have to accept that I don’t know how to read these things. In any case, It sounds good. There. I’ve gone fully over the line. I’m a subjectivist non-audiophile.

 

[12B4A]

I think it's safe to say it's a solid product in its current form as an quasi-omnipolar mono audio source. Its stereo performance will be the real story behind the hype curtain.

 

[Palladium]

Well, I guess I have to accept that I don’t know how to read these things. In any case, It sounds good. There. I’ve gone fully over the line. I’m a subjectivist non-audiophile.

Why wouldn't it sound good? You can usually trust companies with real engineering chops like Apple to deliver the hype. If they can make a killer SoC like the A11 and top tier image/video processing, then audio should be a child's play to them.

Subjectivists preferred boutique outfits like Schitt, not so much.

 

[RayDunzl]

Well, I guess I have to accept that I don’t know how to read these things. In any case, It sounds good. There. I’ve gone fully over the line. I’m a subjectivist non-audiophile.

You're a Musician.

 

[Phelonious Ponk]

Reality is setting in. I set the HomePod on my tv stand, between the JBLs, toggled between them from Apple Music, then turned the SVS sub up to match the balance of bass the pod is emulating. Audiophool hangover causes me to run it at a fraction of what normal people would, and the pod, evidently, is aimed at normal people. The pod doesn’t come very close, IMO. The bass is...boomy?...not exactly, because it’s still well-defined, but it’s not quite right, either. And the mids and highs don’t have the clarity and detail of the JBLs, and bear in mind, while these LSR 305s are really good small recording monitors, they are the world’s cheapest really good small recording monitors. Will two pods bridge the gap? Maybe. I was running it at 100%. Two won’t be working nearly as hard, and the excursion of the bass won’t be dampened as much by the software. Could make all the difference in the world. Will that sharpen the clarity of the mids and highs? Probably not.

With all of that said, it still sounds better off axis than the JBLs. It doesn’t matter where I put it or where I’m sitting. It sounds the same as long as I can see it. They’ve turned the room into the sweet spot, with a $350 speaker. That’s an incredible achievement. I hope Apple, or somebody, does this on a larger scale. And it is, by a long shot, the best speaker I’ve heard in its category. I’ve had the $250 Klipsch (Sent it back. Twice. Long story). And I bought my daughter the $250 Harman Kardon. This is a lot more than $100 better than those, and all the others I’ve auditioned, including more expensive models like the B&Ws. And it takes dictation.


I still may buy a second one, depending on Apple’s return policy (can’t say I’ve ever wanted to return anything to Apple.) But in a place the size of mine, I really only see the need for two if they’re going to replace my main speakers. In the meantime, I think I’ll place it for whole house walking around listening, and pick up my iPad, switch to the JBLs when I’m headed to the living room to sit.

 

[Phelonious Ponk]

You're a Musician.

Yeah, but I do enjoy watching you science perfessers ramble on . Every once in a while I understand what you’re talking about, the rest of the time I can easily imagine that what you’re saying supports what I’d prefer to believe. My expectation bias is highly refined.