--- title: "Noise Floor and Hiss: Why It Happens and Fixes" description: "Why a good amplifier can hiss with sensitive IEMs, why SNR specs do not predict it, and the four fixes that work, with the arithmetic behind each." url: "https://theaudiostuff.com/glossary/noise-floor/" type: "website" author: "Jakub Charkiewicz" --- > Markdown rendering of https://theaudiostuff.com/glossary/noise-floor/ > Canonical URL to cite: https://theaudiostuff.com/glossary/noise-floor/ > Generated from the published page. Text and links only; see the HTML for layout. Breadcrumb: [Home](https://theaudiostuff.com/) > [Glossary](https://theaudiostuff.com/glossary/) > Noise Floor # Noise Floor: Why Sensitive Headphones Hiss The noise floor is whatever a component puts out when there is no signal: hiss, hum and residual circuit noise. Whether you hear it has almost nothing to do with how good the amplifier is and almost everything to do with how sensitive the headphone is, because a sensitive transducer turns a few microvolts of residual noise into audible sound pressure. ## Work it out for your gear [Bit Depth Calculator: Digital Volume Loss](https://theaudiostuff.com/tools/bit-depth-calculator/) Find out how many bits of audio resolution you lose when lowering digital volume in Windows or macOS. Shows effective bit depth and noise floor rise. ## Where the hiss comes from Every active circuit generates noise. Some of it is unavoidable physics: thermal noise in resistors, shot noise in semiconductors. The rest is design, layout and power supply. Added together and measured at the output with the input silent, that residue is the [noise floor](https://theaudiostuff.com/glossary/#noise-floor). An amplifier's [gain](https://theaudiostuff.com/glossary/#gain) acts on its own input noise as well as on the music, which is the part people find counter-intuitive. Turning gain up raises the noise floor at the output by the same number of decibels it raises the signal, so a powerful amplifier set to high gain and then turned down at the volume control is the worst possible partner for an efficient headphone. The noise was amplified before the volume control ever saw it. ## Why the specification does not tell you Signal-to-noise ratio and [dynamic range](https://theaudiostuff.com/glossary/#dynamic-range) are quoted against the component's maximum output. A headphone amplifier rated at 123 dB [SNR](https://theaudiostuff.com/glossary/#snr) is telling you how far its noise sits below full output, and full output is a level nobody listens at. What you actually hear is the noise relative to your listening level, which may be 30 or 40 dB lower, and the gap closes by exactly that much. This is why two amplifiers with the same headline number can behave completely differently with the same [IEM](https://theaudiostuff.com/glossary/iem/), and why the useful figure is output noise in microvolts, which far fewer manufacturers publish. ## The arithmetic, worked twice Take an amplifier with 3 microvolts of residual output noise, which is a respectable figure. Now take a very sensitive in-ear monitor, one that produces 120 dB [SPL](https://theaudiostuff.com/glossary/#spl) for one volt. To reach a normal listening level of 80 dB SPL it needs 40 dB less than a volt, which is 10 millivolts. The noise sits 3 microvolts below that, or about 70 dB down, so the hiss arrives at roughly 10 dB SPL. In a sealed in-ear that blocks the room, that is audible in a quiet space, and it is exactly the complaint people describe as "my expensive amp hisses". Now take a [planar magnetic](https://theaudiostuff.com/glossary/planar-magnetic/) headphone that produces 96 dB SPL for one volt. The same 80 dB SPL now needs about 160 millivolts, the same 3 microvolts of noise is about 94 dB down, and the hiss lands somewhere around minus 14 dB SPL. It does not exist as far as your ear is concerned. Same amplifier, same noise, one audible result and one inaudible one. The variable was the headphone. ## The four fixes, in the order to try them Use the low gain setting. This is free, it costs nothing but voltage swing you were not using, and on most desktop amplifiers it drops the output noise by the difference between the two gain figures. Stop attenuating in software. Cutting digital volume by 30 dB leaves the analogue noise floor exactly where it was while throwing the signal down toward it, and it discards [resolution](https://theaudiostuff.com/glossary/#resolution) at the same time. Keep the operating system and player at or near full, and control level in the analogue domain. The calculator below shows what a given digital cut costs in effective [bit depth](https://theaudiostuff.com/glossary/bit-depth/) and how far it lifts the noise floor relative to the music. Add series resistance, deliberately. An inline attenuator drops both the signal and the amplifier's noise before they reach a sensitive in-ear, which fixes hiss without touching anything upstream. The cost is a change in the source [impedance](https://theaudiostuff.com/glossary/impedance/) the headphone sees, so it interacts with the driver's impedance curve and is a poor idea with multi-driver in-ears whose impedance swings widely. Match the source to the headphone rather than to a specification. A modest, quiet output drives a sensitive in-ear better than a high-power desktop amplifier does, and the desktop amplifier is the right answer for the demanding planar the in-ear is not. This is the whole reason the catalogue tags noise floor as a property worth recording. ## When it is not hiss Hiss is broadband and steady. If what you hear is a low, pitched drone at the mains frequency or its harmonics, that is [hum](https://theaudiostuff.com/glossary/#hum) and it has a different cause and a different fix. If it is a faint chirping or a rhythmic tick that changes when you move a phone or plug in a charger, it is interference reaching the analogue stage, and the answer is usually isolation or a different cable route rather than a different amplifier. If the noise only appears when music is playing and scales with it, it is not the noise floor at all. That is distortion or a bad digital source, and it needs to be traced rather than attenuated. ## What to listen for when auditioning Plug in, select the source, and do not press play. Turn the volume to where you would actually listen and hold still. Then compare the same silence at the same listening level on a second amplifier, not at the same volume-knob position, which means nothing across devices. That test takes thirty seconds and it settles a question that specifications will not. ## Reviewed gear that shows this 5 reviews in the catalogue. Every score is anchored to the same published reference list. - [Synergistic Research PowerCell 14 9.2/10 · $14,495](https://theaudiostuff.com/reviews/synergistic-research-powercell-14/) - [Monosaudio AS-EU08 + SP-USB 8.0/10 · From $200](https://theaudiostuff.com/reviews/as-eu08-sp-usb/) - [Taga Harmony PF-1000LPS 8.0/10 · $445](https://theaudiostuff.com/reviews/taga-harmony-pf-1000lps/) - [DH Labs, iFi, Vera-Fi HiFi Tweaks Roundup 7.8/10 · Varies](https://theaudiostuff.com/reviews/audiophile-tweaks-that-work/) - [Synergistic Research Network Router UEF 7.8/10 · From $2,995](https://theaudiostuff.com/reviews/synergistic-research-network-router-uef/) ## Keep reading - [Best Headphone Amplifiers](https://theaudiostuff.com/guides/best-headphone-amplifiers/) - [Full audio glossary](https://theaudiostuff.com/glossary/) - [How we test and score](https://theaudiostuff.com/about/#methodology-title)