The main types of noise in audio recording
In audio recording, noise usually falls into a few common types: hiss, hum, buzz, clicks and crackle, radio-frequency (RF) interference, room noise, handling noise and quantization noise. Each has a different source, so each has a different fix.
Hiss
Hiss is broadband noise from the electronics in preamps, microphones and cables. Every electronic circuit makes some. Microphone self-noise is specified in dBA. Under about 10 dBA is very quiet, and around 20 dBA is typical of budget condenser mics.
Hiss often gets worse when gain is low and the signal is boosted later. Raise the gain at the source instead.
Hum and buzz
Hum is a steady tone at the mains frequency: 50 Hz in the UK and EU, and 60 Hz in the US. Buzz is hum with harmonics, which makes it sound harsher. Hum usually comes from ground loops or unbalanced cables, although nearby transformers can cause it too. For the steps to fix it, see how to fix static, buzz and digital noise from an audio interface.
Clicks, crackle and pops
These are digital dropouts. They often come from a small buffer size, or from driver or clock settings. Raising the buffer, matching sample rates and using the right driver often clears them.
RF interference
Radio and phone signals can be picked up by cables. A mobile phone can produce a rhythmic “dit-dit” buzz. Move phones away from the setup, and check cables for damage.
Room and handling noise
Room noise comes from air conditioning, traffic, computer fans and reflections from the room. Handling noise is mechanical noise that travels through the mic stand or the mic body. A shock mount reduces it.
Quantization noise and dynamic range
Quantization noise is the small error introduced when an analog signal is converted to digital at a given bit depth. Dither can mask it when you reduce bit depth, for example from 24-bit to 16-bit.
Theoretically, 16-bit audio has a dynamic range of about 96 dB, and 24-bit has about 144 dB. Real converters usually reach about 110 to 125 dB. Using 24-bit doesn’t make a noisy room quieter, because the room’s noise floor sets the limit.
Colors of noise
“Colors of noise” describe how a noise signal’s power is spread across frequencies. They describe a signal type, not a recording problem. You’ll also see these names used for sleep and relaxation sounds, but this article covers them only as signal types.
- White noise has equal power per hertz. It sounds bright and hissy.
- Pink noise has equal power per octave. It falls about 3 dB per octave and sounds balanced to the ear. It’s the standard signal for calibrating monitors and measuring rooms.
- Brown (red) noise falls about 6 dB per octave and sounds like a deep rumble. The name comes from Brownian motion, not the colour.
- Blue noise rises 3 dB per octave, and violet noise rises 6 dB per octave.
- Grey noise follows an equal-loudness curve, so it sounds equally loud across frequencies.
What you can and can’t remove
Noise reduction tools can reduce noise, but they usually can’t remove it all without audible artifacts. A cleaner recording starts with the right gain, grounding and cabling.
FAQ
What does noise suppression mean?
Noise suppression reduces background noise by processing the signal. It can change the sound of the recording, so use it as cleanup, and fix the source where you can.
What is digital noise?
In audio work, “digital noise” usually means clicks, crackle and dropouts that come from the computer, the USB connection or the driver. Quantization noise is a separate, much quieter effect of bit depth.