- Joined
- Mar 30, 2019
- Messages
- 1,172
- Likes
- 2,044
Hi! I published my review of the Google Nest Audio today. You can read that here. It's written for a more mainstream tech audience, but most of the same information is there, so I'll try to keep this post relatively light on the non-measurement-related text.
As I note in my review, when Google gave journalists a more in-depth look at the speaker after its announcement, I felt it deserved a deeper look than a simple quick impression of the sound quality.
The presentation had the usual phrases along the lines of "deep lows and crystal clear highs" and how the speaker will let you hear music "aS tHe ArTiSt InTenDeD," the most annoying phrase in audio marketing. But then the device's product manager said, unlike the original Google Home, this speaker had been designed via "double-blind testing." That made my ears perk up -- that phrase carries some weight around here.
He also talked about maximizing dynamics, increasing thermal dissipation, waveguides, dispersion, minimizing resonances, and internal cabinet volume. All from a top-level overview tailored to mainstream tech coverage, but still, just hearing those things made me feel Google had put some real effort into this design.
And it shows. Here's the spin as measured at 62 percent volume at a distance of 2 ft/0.6m via Bluetooth:
View attachment 86234
I mean, wow. This is a $99 speaker -- for each unit, but still; we've seen much worse spins from speakers far more expensive. The only 'real' issue here is the high-Q dip at 9 kHz, and it's largely balanced out off-axis. Below this, directivity control is excellent, with no visible crossover dip and a listening window that is ±1.5dB from 200Hz to 8kHz. There are no obvious significant resonances. It's quite wide directivity, too.
Someone buy the engineers a pizza.
The response does drop precipitously after 13kHz. I should note that as I used Bluetooth for these tests, I wasn't sure of the sample rate being used, so that might affect the highest frequencies slightly (I got slightly different results for 44kHz vs 48kHz, for example). But this should not be an audible issue for most people. I can hear up to 19kHz fine and though maybe the speakers lacked a little 'air,' at this point that might just be the bias of knowing the measurements.
The elevated bass is not a problem; it is due to some purposeful loudness compensation, and it is subjectively appreciated. At this SPL (admittedly not very loud) you're getting useful in-room bass down to the 40s. More on this later.
I think any way you cut it, this is pretty darn good performance given the price and size. I should also note the speaker is a lot smaller than I had assumed from photos -- its face smaller than a paperback book. It's 6.9 x 4.9 x 3.1 inches (17.5 x 12.4 x 7.8 cm), which makes the unit even smaller than the iLoud Micro Monitor (7.1 x 5.3 x 3.5 in / 18 x 13.5 x 8.9 cm). The woofer is just 3 in (75mm) and the tweeter is 3/4" (19mm).
Here's an image provided by Google of the unit without its cover:
View attachment 86252
You can see that unlike some of the smart speakers out there, this is a pretty traditional design. A woofer, a tweeter, and a waveguide in a sealed enclosure. No fancy weird drivers and layouts, just the good ol' basics. Not that I mind more inventive designs, but only if they contribute something useful.
Let's look at horizontal directivity. First the ER curves:
View attachment 86258
Again, 9kHz and above 13 kHz aside, it doesn't get much better than that.
Here are the raw horizontal curves.
View attachment 86265
This is extremely good for the most part. Among the best I've seen below 8 kHz, frankly. To be fair, having a small woofer for easier integration with the tweeter helps, but still: $99 speaker.
Note the bumps below 500Hz were artifacts of an impromptu measurement setup; I just moved so I have not yet figured out my permanent measurement setup. These bumps are not present in several other measurements of the speaker.
Just to make sure though, I performed a few measurements with the speaker about 8 feet up, further from reflecting surfaces other than some leaves from the trees. Here is one such measurement with a 9ms gate. Note I might've been slightly off axis here as the speaker was too high up for me to confirm alignment, but the point was confirming the lack of resonances below 1kHz.
View attachment 86283
Back to directivity, here are the horizontals normalized to the on-axis.
View attachment 86268
No significant anomalies other than the obvious one.
And the normalized polar map telling us the same thing:
View attachment 86280
The speaker is also well controlled within its vertical listening window, though it is better below the tweeter axis than above it. I feel this should be the other way around, as this is the type of speaker you are more likely to hear while standing up. ± 0/5/15 vertical:
View attachment 86284
Ceiling and floor reflections are pretty typical for a 2-way, which is to say fairly messy:
View attachment 86285
Which you can see as well in the polar map:
View attachment 86286
Luckily evidence suggests that if the verticals don't show up much in the ER and PIR curve, they shouldn't hurt much in terms of tonality.
Lastly, we look at the woofer's behavior at different SPLs. The top curve is 90% volume, the bottom curve is 30%. White is the 62 percent volume the speakers were tested at, which equates to about 71 dB at 1m. I've applied baffle step compensation to each of the measurements, although note this will necessarily be a little bit off; I could not get as close to the woofer as ideal because I couldn't take off the fabric cover. Actual results might be slightly different, but this is good enough for this evaluation:
View attachment 86245I have no idea what the notch at 150Hz is about. The Nest Audio does not have built-in room EQ like the Google Home Max, so it's not trying to fix a room node or anything. Perhaps it is a notch to minimize distortion artifacts that might arise at high SPLs.
View attachment 86297
Some of you know I usually care little about distortion if it's not audible in a pejorative way. In this case, I was a little surprised it wasn't audible at high SPLs, given how small the thing is; other small speakers I've tested tend to show more signs of strain at high volumes.
Granted, I'm not sure how useful distortion measurements of the woofer alone are, but at the very least we can see Google took care to make sure distortion does not rise above the fundamental at high SPLs, as many cheap speakers do. Here's the woofer at the SPL level of the spin:
View attachment 86298
(Note that I could note apply baffle step compensation to distortion, hence the droop)
And here it is at 90:
View attachment 86299
Again, still not crazy loud at 90%. I believe this translates to about 84-85 dB @ 1m. But still, distortion in the bass remains well below the fundamental.
So there you have it. It's a small speaker and it won't get crazy loud or crazy low, but what's there is designed very well considering the low price and design constraints -- not to mention all the other features packed into the speaker. Easy peasy recommendation if you don't need high SPLs and can live without the option for a sub (Amazon added a wireless sub for the echo, so who knows, maybe Google will do the same).
Quick question: were you able to figure out the crossover frequency?
