• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Apple HomePod Review (Smart Speaker)

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
45,717
Likes
254,141
Location
Seattle Area
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:

Apple HomePod Review Smart Speaker.jpg


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:

Apple Homepod Frequency Response measurements smart speaker.png


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:

Apple Homepod early window Frequency Response measurements smart speaker.png


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:

Apple Homepod Predicted in-room Frequency Response measurements smart speaker.png


Horizontal beam width tells us what we already know from its physical construction and measurements above:

Apple Homepod Horizontal Beamwidth  measurements smart speaker.png


The directivity plot is so unusual because of it:

Apple Homepod Horizontal Directivity measurements smart speaker.png


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:

Apple Homepod Vertical Directivity measurements smart speaker.png


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:

Apple Homepod Equalization EQ smart speaker.png


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/
 

Attachments

  • Apple HomePod frequency response.zip
    44.2 KB · Views: 164
Note: Rating not meant for omnidirectional speakers. Also, measurement resolution is roughly 1/2 that from normal (due to measurement length issue explained by Amir).


Preference Rating
SCORE: 5.0
SCORE w/ sub: 5.6

Frequency response: +/- 14.7dB 80Hz-20kHz ; +/- 5.5dB 80Hz-15kHz

Still on a houseboat in Lake Powell, so graphs to come when I have better internet speed.
 
Last edited:
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!

LMAO. Man, I remember when that happened to me and I was banging my head against the wall trying to understand why I had 6 samples (out of 300) that kept failing the "Ambient Noise Test", repeatedly. My test time went from 1.5 hours to literally all night. Totally there with you, even though our issues were completely different. But I'm glad you mentioned the issue for your test because that's something I'll have to watch out for.

Thanks for posting the data. I'm sitting on a couple BT speaker reviews that I won't be publishing for a bit but I do have the Homepod mini data so it'll be cool to compare the two. :)
 
Something interesting of note:

The HomePod does something similar to a surround upmixer, the raw music is played forward, but it extracts the instrumentals and plays them reward, increasing the ambiance/soundstage without muddying the vocals (but of course instrument imaging may not be as good).

The HomePod can auto-calibrate by detecting boundaries (which is how it knows where to send the raw music and where to send the ambient version), and I believe it also adjusts the bass based on boundary distance.

I was going to suggest that maybe since Amir is measuring it in the middle on his garage, that is why the bass is boosted, but if it is still bass heavy when he is listening on his computer desk then I guess not (I believe it takes 30sec to recalibrate).
 
Last edited:
Nice!


Don't forget @napilopez measured the Google Nest Audio here as well which is a good one to compare against as well:
https://www.audiosciencereview.com/...-nest-audio-spinorama-and-measurements.16464/
As are Napier's measurements of Sonos speakers (too hard to dig them up on the phone). We have a handful of these lifestyle products to start with.
Unfortunately he has 120 dB vertical scale which makes the measurements look flat.
Early days of ASR. Long way since then.
 
Note: Rating not meant for omnidirectional speakers.

Preference Rating
SCORE: 5.0
SCORE w/ sub: 5.6

Frequency response: +/- 14.7dB 80Hz-20kHz
Truly beyond baffled by the 5.0 pref rating.
 
While Apple products might try to be 'idiot proof', and suitable for seniors, it is also in many ways non-intuitive and non-plug and play. At least, lost Apple products can be recovered relatively easily.
 
Thanks for testing this Amir. Sonos and echo speakers and Bluetooth speakers are the most popular speakers out there right now so anything you can do to bring people looking to compare them to traditional speakers, to your site to see what the real difference is, is great for the community and budding enthusiasts.

The Homepod is, IMHO, the worst example of Apple hubris. This thing is literally a paperweight of you don't have either an IPhone or apple music subscription. I am confident there are people that bought this thing that could not actually make any sounds come.out of it.
 
Thanks for the interesting review. I think this makes a pretty nice speaker for a kitchen or workshop where you’re moving around off axis and hands free control is advantageous (I have one in the kitchen). In more “serious” listening settings, not so much.

Of course, being deep in the Apple ecosystem is basically required to use this speaker, which makes it a poor choice for Android/Windows users.
 
Omg you r so biased!!! Microsoft guy grrrrr! :mad:

Jk! Amir thank you for the review. :) I’ve been waiting to see your take on this. Results are underwhelming, as I expected. I wonder if one day we’ll get to see your take on Airpods Max…?
 
That was an entertaining read this intro... I so relate to this... Like how much time you spend, why is my freakin password not working here, Why do I have to reset and redo... this highly frustrating, redundant error message, never exactly the same but always the same of the Internet of things. Always in the name of security, but is it really? About security?
 
Unfortunately he has 120 dB vertical scale which makes the measurements look flat.
Even then it looks flatter than the Klippel, I am a bit curious in this case as the apple homepod is definitely aiming at the in room measurement and boosts frequencies accordingly, seems like some active dsp is going on also? In such case it seems the back calculated anechoic measurement being irrelevant and measuring in room represents the actual performance more?
 
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???

Maybe it just was making sure that it was you and not some intruder with a smart gadget pretending to be an Apple speaker... trying to get all that good stuff you keep on your phone? :)
 
  • Like
Reactions: trl
Even then it looks flatter than the Klippel, I am a bit curious in this case as the apple homepod is definitely aiming at the in room measurement and boosts frequencies accordingly, seems like some active dsp is going on also? In such case it seems the back calculated anechoic measurement being irrelevant and measuring in room represents the actual performance more?
Yeah, I wonder how the active DSP plays into these measurements.
 
Back
Top Bottom