This is a review and detailed measurements of the Apple HomePod smart speaker. It is on kind loan from a member and costs US $299. Apple recently discontinued them but I understand a few are still available.
There is no mistake that this is a product from Apple with impeccable look and feel:
View attachment 133255
The only way to communicate with the HomePod is using Airplay and Wifi. There are no inputs for anything else nor does it work with Bluetooth. I was expecting first class installation procedure but went through hell and back. You are supposed to tap your iphone on it to autoconfigure. I did and was pleased to see the pop up but it immediately demanded that I set up 2-step verification for my Apple account. What the heck? What does that have to do with installing a smart speaker???
It proceeded to send me to some other place on the phone where I was told once again my account was locked. Why or why? If you don't use the account for a while Apple just locks it on you? So then had to go through multiple steps of getting the account unlocked and then configure 2-step verification. By then some time had passed and I used the iPhone for my Bluetooth testing. Come back to the HomePod and tap the phone. Nothing happens. I tap it every way I can, no dice. Google and read everything there is to read about this and nothing comes up. The "home" app doesn't see it and won't manually configure it either. Nothing. Two days go by and I get the idea that maybe it is using NFC for autoconfig. Going to search for that and realize I had turned off Bluetooth. On a hunch I turn that on and bam! the pop up works and it autoconfigures. Amazing. This would be bad for any company but Apple? Not even put it in the FAQ that Bluetooth has to enabled?
Once there, the real challenge started which was measuring a speaker like this with Klippel NFS. Normal tests are run synchronous with Klippel generating the sweep and measuring with precision what comes back. That requires line or speaker level input, neither one of which we have. Fortunately Klippel has an way to solve this problem with two microphones (one for measurement and one for time). So I get a second mic, fish it through Klippel and start the first scan. Three hours later it is still going! It should have finished in 2.5 hours. I look and it has reconfigured itself to scan 5 times as long to the tune of 7 hours!
I start over after tweaking thing. Another 3 hours into 2:00am in the morning and the same result. More tweaking and more 3 hour cycles. So I ditch the HomePod and start with a Bluetooth speaker. That created its own problems of how to send the chirp data to it but got that working. So went back to HomePod, 3 more hours and same problem!
I realized that the issue would always appear near the end of the 2.5 hour scan. It would almost be done and then it would go crazy. I play the clip manually and realize first chirp was louder than the rest! The darn HomePod was getting too warm or something and reducing power/changing frequency response. This in turn would confuse Klippel software sending it into a tailspin.
I try to shorten the length of the scan and hit a new bug where no matter what I set it to, Klippel defaults to the same long scan. Frustrated, I blow the whole project, start over (at fair bit of pain due to having to reinitialize everything) and this time I am able to set a shorter scan. And this one worked! Results are below.
This is an unusual speaker in that it has an up firing woofer and then an array of seven tweeters all around the bass at the curvature of the unit. These interfere with each other and produce a massively complex sound field as frequencies go up. This in turn requires a ton of samples in NFS so that you are not limited by aliasing. Alas, due to HomePod overheating, I cannot run may samples. So what you see below is accurate to about 8 kHz. Above that, there is some error but on-axis response shows surprisingly good agreement so I think we are fine.
NOTE: I set the reference axis at the bottom of the speaker where the tweeters are. Setting it higher makes no sense since there is no driver there. In use, this speaker would be on a shelve with tweeters reflecting up from it. Same thing happened in the measurements in that the stand for Klippel which small for normal speakers was larger than the homepod. I thought this better represented how it is used so I went with it.
Apple HomePod Measurements
Alright, let's start with our standard frequency response graphs and "spin" data:
View attachment 133256
We instantly see the heavy bass emphasis and somewhat uneven response. Due to high frequency drivers firing in each way, on-axis actually shows less energy that total sound power which is the opposite of how normal speakers behave. Your actual "on-axis" may be more like Sound Power depending on whether the rear firing drivers reflect back into the front. The omni directional aspect of the drivers naturally generates very symmetrical off-axis response and indeed, off-axis is the same as on-axis:
View attachment 133257
Predicted in-room response doesn't quite apply to this speaker but I suspect it is speaking the truth when it says there is too much bass here:
View attachment 133258
Horizontal beam width tells us what we already know from its physical construction and measurements above:
View attachment 133259
The directivity plot is so unusual because of it:
View attachment 133260
Vertically we have a puzzle that takes some work to decipher due to reflections from measurement stand looking down at the speaker, and blockage looking up:
View attachment 133261
I don't have other measurements for you as I lost my on-axis in-room distortion graph when I restarted the project. Can generate them again if there is time and interest.
Apple HomePod Listening Tests
I started with my usual female vocals and sound there was pretty good. Even at max volume there is no clipping or any sign of overt distortion. Oh, for placement I put the speaker on my equipment stand where I listen to near-field speakers. Listening distance was about 1 meter/3.3 feet. Gradually I started to get annoyed by the level of bass. It was always present no matter what I was playing. So I reached into my equalization tool and cleaned up the response some:
View attachment 133262
The result was much, much cleaner sound Gone was that constant boominess. The EQ is likely not correct in that room modes interfere so more exactly work needs to be done in situ. Taking away the extra bass seemed to accentuate the highs a bit. Measurements didn't show this so don't know why this was the case.
The omni aspects of the sound was nice as I moved left and right a ton and tonality remained the same which was very good.
When it was said that done, it was a mono speaker that could play somewhat loud without distortion. Eh. Nothing exciting. A pair of active speakers would sound so much better. So in the larger context for our kind or use, this is a bust.
Conclusions
The Homepod obviously has benefited from excellence in design. I wish they had used their DSP though to provide different target responses. I get that bass sells but give the rest of us some tuning. Oh, I tired "SoundCheck" but it did nothing that I could detect in listening tests.
At the end, there was nothing exciting for me here. I could not listen to the Homepod without EQ. With EQ it was fine but I wasn't sure what I was getting at the end.
Overall I can't recommend the Apple HomePod. Go ahead and complain that I am a Microsoft guy and don't like Apple product. I will point you right back at the above measurements.
------------
As always, questions, comments, recommendations, etc. are welcome.
Appreciate any donations using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
Oh man, this sound alike an absolute pain, thank you for this Amir. It's definitely hard to contextualize an omnidirectional speaker like this with a variable response but it strikes me as a goodish speaker that likely sacrifices some potential sound quality in order to achieve the omnidirectional gimmick. For it's intended use though it can be a pretty good gimmick, especially for folks with an open floor plan and a single smart speaker
At least with sonos you can disactivate trueplay and extra processing. Hopefully the HomePort didn't vary it's response too much during testing. The Sonos Roam was constantly changing while I was measured outdoors until I turned off trueplay.
