Hiss Calculator

The other half of amp matching, and the half nobody calculates. A powerful amp is useless with sensitive IEMs if you can hear it hissing between tracks. Enter your headphone's specs and your amp's noise figure to find out whether that background hiss will be audible before you buy.

How to read this

The figure is the sound pressure level of your amplifier's own noise, delivered into your headphones with no music playing. Because a sealed IEM blocks much of the room, your effective noise floor while wearing one can be as low as 15–20 dB SPL, so hiss around that level starts intruding on quiet passages. Over-ear headphones leak more room noise in and are usually far less sensitive, which is why hiss complaints come almost entirely from IEM owners.

Typical output noise figures: a well-measured desktop amp sits at 1–2 µV, a good dongle around 1–3 µV, an older DAP or a high-gain setting 5–20 µV, and tube amps can run into the hundreds. Note that gain matters as much as the amp: the same amp on high gain multiplies its own noise, which is why a low-gain switch is often the whole fix.

Where the line falls, across the whole database

Take 20 dB SPL as the point where hiss stops being theoretical — roughly the residual noise you hear inside a sealed in-ear in a quiet room. Run every model here against a range of amplifier noise floors and the picture is lopsided:

5 of the 222 in-ears cross that line at two microvolts. At five it is 42, at ten it is 138 — more than half. Over the same range the 345 full-size models here go from 1 to 44, which is the whole reason this is an in-ear problem and not a headphone problem.

The models that hear everything

Sensitivity and impedance together decide it: a high per-milliwatt figure on a low impedance turns a few microvolts into audible noise. These are the eight worst cases in this database, with the hiss level a two-microvolt and a ten-microvolt source would put into them. Every figure comes from the maker's own sensitivity and impedance.

The Campfire Andromeda's reputation as a hiss detector is not folklore — it falls out of the maker's own two numbers. If you own one of these, the amplifier's noise floor matters more than its power, and a source that drives a Susvara is not automatically a source that is quiet enough for this.

The reason this calculator often cannot be used

83 of the 280 sources in this database publish an output noise figure. The other 197 publish none, so there is nothing to put in the box above. The median of the ones that do is 2 µV; 11 are at or under one microvolt; the quietest published figure here is 0.2 µV (Topping A90) and the highest is 25.18 µV (Schiit Mjolnir 3). That spread says less than it looks, though: the highest figure here belongs to a maker that publishes an absolute number at all. An amplifier that publishes nothing could be worse and you would have no way of knowing.

Those figures are also less comparable than they look. Some makers state A-weighting next to the number, some state a measurement bandwidth, and many state neither — and an unweighted figure and an A-weighted one for the same amplifier are not the same number. Many do not say which gain setting it was measured at either, which for a noise floor is the setting that matters most. What makers publish, counted.

Two practical notes

First, if an amp only quotes a signal-to-noise ratio rather than an absolute noise figure, you cannot use it here directly — SNR is relative to full output, so a powerful amp can post an excellent SNR and still hiss badly into IEMs. Second, an impedance adapter or L-pad drops the hiss and the music equally, so by itself it buys nothing — it only helps once you turn the volume back up, which is why it needs an amp with volume to spare and a noise floor that comes from the output stage rather than from the gain setting.

Every published noise figure, quietest first

All 83 of the 280 sources here whose own documentation states an output noise figure, in order. The two right-hand columns are the part the spec sheet cannot give you: where that noise lands in the most sensitive in-ear in this database — the Shure SE846 at 134 dB/V and 9 Ω — and how many of the 222 in-ears here stay under the 20 dB SPL this site treats as the point where a quiet room stops hiding it.

11 of the 83 are at or below one microvolt. The loudest published figure belongs to the Schiit Mjolnir 3 at 25.18 µV, which puts its own noise at +42 dB SPL in the Shure SE846 and leaves it inaudible in 18 of the 222. The quietest is the Topping A90 at 0.2 µV.

The Shure SE846 does not have quite the highest voltage sensitivity in this database. It has the highest among the in-ears whose maker said which unit the figure is in, and the distinction is not pedantry: 2 models above it on paper (Simgot EA500, Tanchjim EA FORCE) carry a decibel figure with no reference unit printed anywhere, so read per milliwatt they would imply a sensitivity no other in-ear here comes near. A column of hiss estimates cannot rest on that.

Those two models cost the whole list dearly, and the size of it is worth seeing. Counting all 222 in-ears, 5 sources are quiet enough to vanish in every one. Counting only the 145 whose maker stated the unit, 33 are. The hardest bar in this database is set by two numbers that may not mean what they appear to mean, and the gap between 5 and 33 is what that costs anyone trying to choose on published figures.

A high figure here is not a fault and a low one is not a recommendation: a desktop amplifier with four microvolts of noise is silent on 300 Ω headphones and audible on a 9 Ω in-ear, which is why the last column counts in-ears and not headphones. Nor is a missing figure a good sign — the 197 sources that publish no noise figure at all are simply not on this list, and nothing here should be read as a verdict on them. How many that is, and what they print instead, is its own page.