A NAS runs all the time, which makes it one of the few machines where the idle figure is the only figure that matters. This category also has the best noise disclosure we have found anywhere, and one maker that publishes nothing at all.
We read specification pages and datasheets for eighteen products from five brands, covering two bay, four bay, single bay and an all flash model with no drive bays at all.
Our 7 picks are listed above, with full specifications for each.
Synology publishes the test, not just the number
Here is a Synology footnote in full, from the DS225+ page:
"Noise testing was conducted with the Synology system fully installed with Synology SATA HDDs and in an idle state. Two GRAS 40AE microphones are placed at a 1 m distance from the front and rear of the device."
The page then adds the background noise level, the temperature range and the humidity range. That is a named instrument, a named position, a distance, and the conditions of the room. No appliance maker on this site publishes any of those, and no other NAS maker here publishes a microphone type.
There is one gap, and it is a real one. Synology ships three fan modes, Full-Speed, Cool and Quiet, and never says which mode the figure describes.
One practical note if you go looking for these numbers yourself: they are on the product page specification tab and not in the datasheet PDF. We downloaded seven Synology datasheets and none of them contains a noise figure.
The drives make the noise, not the box
Synology's own range proves it. The DS223 publishes 14.6 dB(A), measured "with Synology SATA SSDs". The DS223j publishes 18.2 dB(A), measured "Fully loaded with Seagate 2TB ST2000VN000 hard drive(s)". Same maker, same 92 mm fan, similar chassis, and a 3.6 dB gap that comes from what is in the bays.
So the first question about any figure in this guide is not how low it is. It is what was fitted when they measured it. A 14.6 dB(A) box with four spinning drives in it is not a 14.6 dB(A) box.
This is also why the Asustor FLASHSTOR is interesting despite publishing no conditions at all. It has no drive bays, only M.2 slots, so there is nothing inside it that spins.
A number below the noise floor
The DS223's footnote states a background noise level of 16.49 to 17.51 dB(A). Its published figure is 14.6 dB(A).
A device cannot be measured directly at a level below the room it sits in, so that number has to have been derived some other way. Synology does not say how. We have recorded it as published, flagged it, and kept it out of the top pick, because a figure you cannot interpret is not a figure you should buy on.
Where the conditions thin out
TerraMaster publishes a background level of 17.3 dB(A) and a 1 m distance, which is better than most. But its condition reads "Using 2 SATA HDDs/SSDs in standby mode", which leaves the drive type unresolved, and it uses that same two drive condition for the four bay F4-424. A four bay machine measured half empty is not the machine you will own. TerraMaster also names an exact drive for its power consumption test and no drive at all for its noise test.
Asustor publishes decibels in marketing sentences rather than specification rows, and quotes the same 19 decibels for a two bay and a four bay in the same series.
UGREEN is the only maker here that names a fan mode, "In quiet mode, noise levels stay between 29 and 34dB". Those are the highest numbers in this guide, which may mean UGREEN is louder, or may mean UGREEN is measuring something the others are not.
QNAP publishes nothing. We checked sixteen current models and found no decibel figure on any specification page. Since QNAP's downloadable datasheet is generated from that same table, there is no fallback.
Synology's disclosure is getting worse, not better
The older Synology footnote names the drive: "Fully loaded with Seagate 2TB ST2000VN000 hard drive(s) in the idle state". The newer one says only "Synology SATA HDDs".
Both generations are live on Synology's site right now, which is how the DS223j can tell you exactly what was in it while the newer DS225+ cannot. If you want a figure you could reproduce, the older models are the ones that let you.
We did not measure any of these machines. Every figure here is the maker's own, quoted as published, with the condition attached wherever the maker attached one.
