Noise floor explained for clean professional audio

Every recording contains a certain amount of unwanted sound. Some of it comes from the room, such as air conditioning, traffic or computer fans. Some is generated by the equipment itself, including microphone electronics, preamps, converters and analogue circuitry. This constant background level is called the noise floor.

Noise floor is usually measured in decibels, with the reference changing according to the recording system. In a digital workstation, levels are generally shown in dBFS, where 0 dBFS is the loudest possible digital level. A quiet recording may show a noise floor around -70 dBFS, while a poorly controlled session might sit at -45 dBFS or higher.

A low noise floor gives an engineer more room to shape a performance. Quiet vocal phrases, sustained guitar notes and natural pauses remain clean when the track is compressed, equalised or lifted in level during mastering. If the background noise is already loud, processing can make the problem much more obvious.

For artists working in Australia, the issue can appear in many practical settings: a bedroom studio in Melbourne, a rehearsal room in Sydney, or a converted garage in Brisbane may all have different acoustic and electrical conditions. Professional recording in Elgin, Illinois, follows the same technical principles, but the right target depends on the source, the room and the final format.

What the noise floor actually measures

The noise floor is the lowest average level present when no intentional signal is being recorded. To measure it, an engineer can leave the input active, ask the performer to remain silent, and examine the waveform or metering in the digital audio workstation. The result reflects the combined noise of the acoustic space, microphone, cable, preamp, converter and monitoring chain.

This measurement should not be confused with a single brief click or isolated hum. A noise floor is a continuing background level, often heard as hiss, buzz, rumble, air movement or low-level electronic interference. Short events such as a chair scrape or passing truck are separate noises, although they may still ruin an otherwise quiet take.

Digital meters can show the noise floor in several ways. Peak meters reveal sudden spikes, while RMS or short-term loudness measurements give a better indication of sustained background sound. A spectrogram can add useful information by showing whether the noise is broad hiss, low-frequency rumble or a narrow electrical tone at 50 Hz and its harmonics.

Why a lower level is usually better

A quieter recording preserves greater usable dynamic range. This matters when a singer moves from a soft verse into a powerful chorus, when a narrator leaves deliberate pauses, or when a string player allows a note to decay naturally. The more distance between the wanted performance and the noise floor, the more confidently an engineer can edit, automate and process the track.

Compression can expose noise because it raises quieter material towards the rest of the programme. Equalisation may do the same, especially when adding high frequencies where microphone hiss is concentrated. Reverb also carries the background into spaces between phrases, creating a hazy or dirty result rather than a controlled sense of depth.

A very low noise floor is not a substitute for a good performance or suitable acoustics. If the microphone is too far away, the engineer may need excessive gain, which increases room sound and equipment noise. Good microphone placement, appropriate preamp gain and a controlled room often produce a better result than simply buying a more expensive converter.

For spoken-word projects, this control is especially important. Audiobooks, narration and voice-over work may contain long pauses, quiet consonants and exposed breaths. A noisy pause can be more distracting than a noisy musical passage because the listener has fewer competing sounds to mask it.

How low should it be?

There is no single number that suits every professional recording. For music, a practical tracking target is often a sustained noise floor around -60 dBFS or lower, with many well-treated studios achieving -70 dBFS or below on suitable sources. These figures are broad working references rather than universal pass marks. The source level, microphone sensitivity and gain structure all affect the measurement.

For an isolated vocal, acoustic guitar or voice-over, a floor below -60 dBFS is commonly considered workable when the performance has healthy peaks well above it. A result closer to -70 or -80 dBFS gives additional safety for heavy editing and dynamics processing. The important figure is the signal-to-noise ratio: a quiet room is useful only when the wanted signal is recorded clearly above the background.

Audiobook and broadcast delivery may have stricter requirements set by a publisher, platform or client. Some audiobook specifications refer to an RMS noise level of -60 dB or lower, along with limits for peaks and overall loudness. The exact delivery standard should be checked before recording, since platform rules can change and different clients may request different technical measurements.

In Australia, a commercial voice-over intended for a Sydney agency, a Melbourne production house or national advertising campaign may need to meet the client’s own technical sheet rather than a general internet standard. Ask for the delivery requirements before the session, particularly if the recording will be used for television, radio, podcast distribution or an online learning platform.

What creates unwanted noise

Room noise is often the first problem. Air conditioners, refrigerators, fluorescent lights, road traffic and building vibration can all enter a microphone. In Australian homes, early-morning birds, barking dogs and passing utes may become part of a take, while coastal locations can add wind through poorly sealed windows. Choosing a quiet recording time can help, but acoustic treatment and isolation provide more dependable control.

Electrical interference has a different character. A ground loop can produce a low hum, while poor shielding may introduce buzz or radio-frequency interference. A balanced connection, correctly earthed equipment and sensible cable routing can reduce these faults. Replacing cables at random is less effective than identifying whether the noise changes when devices are connected, moved or powered from another circuit.

Gain staging is another major factor. If the preamp gain is set too low, the engineer may need to raise the track substantially later, lifting the equipment’s self-noise and any room sound. If it is set too high, loud transients may clip. Recording in 24-bit resolution gives a useful safety margin, so there is no need to chase very hot levels during tracking.

Microphone choice and placement also shape the result. A sensitive condenser may capture detail and low-level ambience that a dynamic microphone rejects. Close placement can improve the ratio between voice and room, although it may increase plosives or proximity effect. A pop filter, stable stand and consistent working distance can often improve clarity more than aggressive noise reduction after recording.

Measuring and fixing the problem

Start by recording a short room tone before the performer begins. Keep the microphone, preamp and input settings exactly as they will be during the take. Watch the meter for a minute and listen with headphones at a sensible level. This reveals whether the issue is steady hiss, intermittent environmental noise, hum or something caused by the monitoring setup.

Use the waveform and a spectrum analyser together. A steady broadband rise across the high frequencies often indicates hiss, while a strong peak at 50 Hz points towards mains hum in Australia and other regions using 50 Hz electricity. Harmonics at 100 Hz, 150 Hz and beyond can indicate a more complex grounding or power problem. Low-frequency energy below the useful range may come from traffic, footfall or mechanical vibration.

Fix the physical source before reaching for a plug-in. Turn off unnecessary fans, move the microphone away from computers, close windows, improve cable separation and check power connections. If the room is reflective, absorptive panels and suitable microphone positioning can reduce the amount of gain needed. A professional facility with a controlled recording environment can save considerable restoration work later.

Noise-reduction software has a legitimate role when used carefully. A denoiser can learn a short noise print and reduce a consistent background, while a high-pass filter can remove unwanted rumble. However, strong processing can create watery artefacts, chirping or an unnatural loss of air. Restoration should preserve the character of the performance rather than make every silent gap unnaturally empty.

Keeping a clean signal through mixing and mastering

A low noise floor at tracking gives the mix engineer more options, but silence should still be edited intelligently. Unwanted noises between phrases can be removed or reduced with clip gain, fades and automation. Abrupt digital silence may sound artificial, so retaining a small amount of consistent room tone can create smoother transitions.

Do not confuse normal ambience with a technical fault. The breath of a room, the mechanical texture of a tape machine or the natural noise of a valve amplifier may contribute to a recording’s identity. The question is whether the sound distracts from the content and whether it becomes more prominent after compression, limiting or loudness normalisation.

Mastering can reveal problems that were difficult to notice in the mix. Raising the overall level brings quiet tails and background hiss forward, particularly in sparse arrangements. Album sequencing also exposes inconsistencies: one song with a much higher floor can sound noticeably dirtier when it follows a quiet track. Decisions about a single release and a full album can therefore involve different editing and mastering priorities, as explained in this mastering cost guide.

Streaming services reduce or increase playback level through normalisation, but they do not remove the noise embedded in the audio. A noisy vocal remains noisy when the listener turns up the volume during a quiet passage. Clean source files, appropriate headroom and controlled dynamics remain important for releases on platforms used by Australian listeners, from independent Bandcamp audiences to major streaming services.

Practical targets for different projects

Music production usually benefits from a quiet tracking chain, but the acceptable floor depends on arrangement density. A full rock production may mask a small amount of hiss, while a solo folk recording, classical performance or intimate singer-songwriter track will expose it. For a lead vocal, aim for a clean signal with a comfortable margin above a floor around -60 dBFS or lower, then judge the result in the context of the mix.

Podcasts and interviews need consistency between speakers and episodes. A guest recorded from a noisy home office can make editing difficult when placed beside a presenter captured in a treated studio. Recording a few seconds of room tone from each participant helps match edits, although it cannot fully repair changing traffic, fans or room reflections.

Use these working practices to keep the background under control:

• Record at 24-bit resolution and leave sensible headroom rather than forcing peaks close to 0 dBFS.
• Capture a minute of room tone and inspect the noise before recording the full performance.
• Keep the microphone close enough for a strong direct signal, while controlling plosives and proximity effect.
• Investigate hum, buzz and rumble at the physical source before applying restoration plug-ins.
• Check the quietest passages after compression, limiting and loudness normalisation.

Professional audio does not require absolute silence. It requires a noise floor low enough that the listener notices the performance, not the recording chain. For a band, solo artist, audiobook narrator, advertiser or podcaster, the best target is the lowest practical level that can be achieved without compromising tone, comfort or workflow.