Learn

Spectrum analyzer glossary

Sixty words you will meet on spectrum analyzers, in audio and in radio, each explained in a few plain sentences with real numbers. Jump to a letter, or type in the box to filter. Where this site has a tool or a guide for a term, a link takes you straight there.

A

A-weighting
A filter curve that copies how the ear hears quiet sounds: it cuts deep bass and the very top while keeping the middle. Levels measured with it are written dB(A). It is the usual choice for noise limits and hearing safety, but it makes bass look much smaller than it is. Try it: Real-Time Analyzer (RTA) →
Aliasing
A false frequency that appears when a signal above half the sample rate is sampled. At 48 kHz, a 30 kHz tone shows up as 18 kHz. Recorders and sound cards filter out such tones before sampling to stop it, so on a good input you rarely see it. Try it: FFT & Sampling Calculator →
Attenuation
Making a signal smaller on purpose, usually stated in dB. A 20 dB attenuator passes one hundredth of the power. On an RF analyzer the input attenuator protects the mixer from strong signals, but each step of attenuation also raises the noise floor by the same amount. Try it: dBm Converter →
Averaging
Combining many spectrum frames into one steadier line. Noise and random wobble shrink while steady tones stay put, so small peaks become easier to read. Exponential averaging follows changes quickly; linear averaging over a set number of frames gives a fixed, repeatable result. Try it: Live Spectrum Analyzer →

B

Bandwidth
The width of a range of frequencies, in Hz. A phone call carries about 3 kHz of bandwidth, an FM station about 200 kHz and a Wi-Fi channel 20 MHz or more. On an analyzer the word also appears in RBW and VBW, which are the widths of its filters. Read: What Is a Spectrum Analyzer? →
Bin
One slice of frequency in an FFT result. The width of each bin is the sample rate divided by the FFT size: 48,000 / 8,192 gives about 5.9 Hz per bin. Smaller bins show finer detail but need a longer piece of sound, so the display reacts more slowly. Try it: FFT & Sampling Calculator →

C

C-weighting
A filter curve that is almost flat across the audio range and only rolls off the very lowest and highest frequencies. Written dB(C). It is used for loud sound, peak levels and judging bass, where A-weighting would hide most of the low end. Try it: Real-Time Analyzer (RTA) →
Centre frequency
The frequency in the middle of the screen on an RF spectrum analyzer. Together with the span it sets what you see: a centre of 433.92 MHz and a span of 2 MHz shows 432.92 to 434.92 MHz. In audio it also names the middle of a filter or an octave band. Read: Spectrum Analyzer Settings →
Clipping
What happens when a signal is pushed past the biggest level a system can hold: the tops of the wave are cut flat. It sounds harsh and adds new harmonics. On a spectrum it shows as a sudden spray of extra peaks; on an oscilloscope the flat tops are easy to see. Try it: Clipping Detector →
Crest factor
How much higher the peaks of a signal are than its average (RMS) level. A sine wave has a crest factor of 1.41, or 3 dB. Speech and music are much spikier, often 12 to 20 dB, which is why they need headroom to avoid clipping. Try it: Online Oscilloscope →

D

dB (decibel)
A way of comparing two levels on a log scale. Every 10 dB is ten times the power, every 3 dB about double, and 6 dB doubles a voltage. A dB value always needs a reference, which is why there are many kinds: dBFS, dBm, dB SPL and more. Read: What Is a Spectrum Analyzer? →
dBFS
Decibels relative to digital full scale. 0 dBFS is the largest value a digital system can store, so every real signal reads below zero: −6 dBFS is half the largest possible wave, and −20 dBFS is one tenth. Any attempt to go above 0 dBFS clips. Try it: Live Spectrum Analyzer →
dBm
Power in decibels compared with one milliwatt. 0 dBm is 1 mW, +30 dBm is 1 W and −90 dBm is one billionth of a milliwatt. RF analyzers show signal strength in dBm; a Wi-Fi signal at your laptop is often between −40 and −80 dBm. Try it: dBm Converter →
Decibel SPL
Sound pressure level in decibels, compared with 20 micropascals, about the quietest sound a young ear can hear. A quiet room is near 30 dB SPL, talking about 60, a loud concert over 100. A phone needs calibration before its readings mean real dB SPL. Try it: Real-Time Analyzer (RTA) →
DFS
Dynamic Frequency Selection: a rule for many 5 GHz Wi-Fi channels that are shared with weather and military radar. A router on a DFS channel must listen for radar first and move away when it hears it, so those channels can take a minute to start and may switch without warning. Try it: WiFi Channel Planner →
DTMF
Dual-Tone Multi-Frequency: the tones a phone keypad makes. Each key plays two tones at once, one from a low group (697 to 941 Hz) and one from a high group (1,209 to 1,633 Hz). On a spectrum each key shows as exactly two peaks. Try it: DTMF Tone Decoder →
Dynamic range
The span between the largest signal a system can handle and the smallest it can still show above its own noise, in dB. 16-bit audio gives about 96 dB. A good analyzer shows signals 70 to 100 dB apart on one screen, far more than most oscilloscopes. Read: Spectrum Analyzer vs Oscilloscope →

E

ENBW
Equivalent noise bandwidth: how much noise one FFT bin collects, compared with a perfect rectangular filter of one bin width. The Hann window has an ENBW of 1.5 bins. It matters when you turn a noise reading per bin into a level per Hz. Try it: Window Function Explorer →
ENOB
Effective number of bits: how many bits of a converter are really useful once its noise and distortion are counted. A 16-bit converter that measures 13.5 ENOB performs like a perfect 13.5-bit one. It is worked out from the converter's SINAD. Try it: ENOB Calculator →

F

FFT
Fast Fourier Transform: a quick way to work out which frequencies are in a short piece of signal and how strong each one is. A spectrum analyzer runs it many times a second on fresh samples. The FFT size sets how many samples each run uses. Read: FFT Explained →
Flat-top window
An FFT window shaped to read the level of a tone exactly, even when the tone falls between two bins. The price is wide peaks, so two close tones blur into one. Use it for level measurements and calibration, and a Hann window for general looking. Try it: Window Function Explorer →
Formant
A band of frequencies boosted by the shape of the mouth and throat. Each vowel has its own formants: in "ah" the first two sit near 700 and 1,100 Hz, in "ee" near 300 and 2,300 Hz. On a spectrum they show as humps over the voice's harmonics. Try it: Voice Formant Analyzer →
Frequency response
How evenly a device passes each frequency, usually drawn as level in dB from 20 Hz to 20 kHz. A flat line means every frequency comes out as loud as it went in. Speakers, mics, headphones and rooms all have their own, and none is perfectly flat. Try it: Frequency Response Test →
Fundamental
The lowest frequency of a repeating sound, which usually sets the pitch you hear. A guitar's open A string has a fundamental of 110 Hz. The harmonics above it, at 220, 330 and 440 Hz and so on, give the instrument its colour. Try it: Harmonics & Overtone Analyzer →

H

Hann window
The most common FFT window. It fades each piece of signal in and out smoothly, which keeps a tone's energy close to its own bins instead of spreading it across the screen. A good default for audio and most general use. Try it: Window Function Explorer →
Harmonic
A frequency at a whole-number multiple of the fundamental: two, three, four times and up. Instruments and voices have many harmonics, and their mix is what makes a trumpet sound different from a flute. Unwanted harmonics from an amplifier are counted as distortion. Try it: Harmonics & Overtone Analyzer →
Hz (hertz)
The unit of frequency: how many times a second something repeats. 1 kHz is 1,000 Hz, 1 MHz a million and 1 GHz a thousand million. People hear from about 20 Hz to 20 kHz, while radio, Wi-Fi and phones use MHz and GHz. Read: Audio Frequency Chart →

I

Impulse response
How a system or a room answers one very short click. It holds everything about the system's sound: the direct sound, the echoes and how fast they die away. Room tools measure it with a sweep or a clap and work out RT60 and the frequency response from it. Try it: RT60 Reverb Time Meter →
Intermodulation
New, unwanted frequencies made when two or more signals pass through something that is not perfectly linear. Tones at 1,000 and 1,100 Hz create products at 900 and 1,200 Hz, and many more. In radio these products can land on another channel and block it. Try it: Harmonics & Intermodulation Calculator →
IP3
Third-order intercept point: a figure that says how well an amplifier or receiver handles strong signals before it makes intermodulation. A higher IP3 is better. Third-order products grow 3 dB for every 1 dB of input, so they rise fast once signals get strong. Try it: Harmonics & Intermodulation Calculator →

L

Leakage
Energy from one tone spilling into neighbouring FFT bins, so a pure tone looks like a wide hill instead of a thin line. It happens because each FFT piece does not hold a whole number of cycles. Window functions reduce leakage. Try it: Window Function Explorer →
Log scale
A scale where each step multiplies instead of adds. On a log frequency axis, 20 to 200 Hz takes the same width as 2 to 20 kHz, which matches how we hear pitch. Decibels are a log scale for level, so tiny and huge signals fit on one screen. Read: How to Read a Spectrum Analyzer →
LRA (loudness range)
Loudness range: how much the loudness of a track changes from its quiet parts to its loud parts, in LU. A pop song may have an LRA of 4 to 6 LU, a film or a classical piece 15 or more. The very quietest and loudest moments are left out so short events do not skew it. Try it: LUFS Loudness Meter →
LUFS
Loudness Units relative to Full Scale: a loudness measure that follows how loud music sounds, not just how big its peaks are. Streaming services turn songs to a target near −14 LUFS, and broadcast in Europe uses −23 LUFS. Each LU is the same size as 1 dB. Try it: LUFS Loudness Meter →

M

Mains hum
A low buzz at the frequency of the power grid, 50 Hz in most of the world and 60 Hz in North America and parts of Asia, often with harmonics at 100, 150 Hz or 120, 180 Hz. It usually comes from a ground loop or a cable running near power wiring. Read: How to Fix Mains Hum →
Marker
A point you place on the spectrum to read its exact frequency and level. Most analyzers can jump a marker to the highest peak and show the difference between two markers, which is the easiest way to measure how far a harmonic sits below its fundamental. Try it: Live Spectrum Analyzer →

N

Noise figure
How much noise an amplifier or receiver adds to a signal, in dB. A perfect device would have 0 dB. Low-noise amplifiers reach below 1 dB. The first stage in a chain matters most, which is why a good preamp goes right at the antenna. Try it: Noise Figure Cascade Calculator →
Noise floor
The lowest level an analyzer can show, set by the noise of its own input and by its settings. It looks like an uneven band along the bottom of the graph. A signal must stand clearly above it to be measured; a narrower RBW or longer FFT lowers it. Try it: Noise Floor & RBW Calculator →
Nyquist frequency
Half the sample rate: the highest frequency a digital system can hold. At 44.1 kHz it is 22.05 kHz, at 48 kHz it is 24 kHz. Anything above it folds back as aliasing, so a spectrum stops at the Nyquist frequency. Try it: FFT & Sampling Calculator →

O

Octave
A doubling of frequency. 100 Hz to 200 Hz is one octave, and so is 1 kHz to 2 kHz. Hearing spans about ten octaves. RTAs split sound into octave or 1/3-octave bands, which get wider in Hz as they go up, just as our sense of pitch does. Try it: Octave Band Calculator →
Overtone
Any frequency in a sound above its fundamental. In most instruments the overtones are harmonics, at whole-number multiples. In bells, drums and some metal bars they are not, which is why those sounds have no clear single pitch. Try it: Harmonics & Overtone Analyzer →

P

Peak hold
A second line on the spectrum that keeps the highest level each frequency has reached. It catches short events the live line would miss, such as a brief radio burst or one loud note. Clear it to start again, or let it slowly fall back. Try it: Live Spectrum Analyzer →
Phase
Where a wave is in its cycle, in degrees: 0° at the start, 180° halfway, 360° back at the start. Two equal tones in phase add up to double; at 180° apart they cancel. Phase is why a mic between two speakers can lose some frequencies. Try it: Fourier Playground →
Phase correlation
A number from −1 to +1 that shows how alike the left and right channels of a stereo signal are. +1 means mono, near 0 means wide stereo, and below zero means parts are out of phase and will cancel when the track is played in mono. Try it: Stereo Phase & Width Meter →
Pink noise
Random noise with equal energy in every octave, so it sounds even and not hissy. On an FFT spectrum it slopes down 3 dB per octave, but on an RTA with octave bands it reads flat. That makes it the standard test signal for rooms and speakers. Try it: Real-Time Analyzer (RTA) →
Polarity
Whether a signal or a speaker pushes or pulls when the input goes positive. A speaker wired the wrong way round has reversed polarity: it pulls when the other pushes, and the bass between them cancels. It is fixed by swapping the two wires on one speaker. Try it: Speaker Polarity Test →

Q

Q factor
How narrow and sharp a peak or a filter is: the centre frequency divided by its width. A Q of 10 at 1 kHz is 100 Hz wide. High-Q peaks ring for a long time; room modes, guitar bodies and EQ filters are all described by their Q. Try it: Resonance Frequency Finder →

R

RBW
Resolution bandwidth: the width of the filter an analyzer uses to measure each point. A narrow RBW separates close signals and lowers the noise floor, but sweeps more slowly. In an FFT analyzer the bin width plays the same part. Read: Spectrum Analyzer Settings →
Reference level
The level at the top line of an RF analyzer's screen, in dBm. Set it a little above the strongest signal. Too high wastes screen space and can hide weak signals in the noise; too low lets strong signals overload the input and make false peaks. Read: Spectrum Analyzer Settings →
Room mode
A bass frequency at which sound bounces between walls and builds up into a loud, ringing peak, or cancels into a dip, depending on where you sit. The size of the room sets these frequencies: a 5 m long room has its first mode near 34 Hz. Try it: Room Mode Calculator →
RT60
Reverberation time: how long it takes a sound to fade by 60 dB after it stops. A small, furnished room is often 0.3 to 0.5 s, a church several seconds. Too long and speech gets muddy; panels and soft furniture bring it down. Try it: RT60 Reverb Time Meter →

S

Sample rate
How many times a second a digital system measures the signal. CDs use 44,100 per second and video 48,000. Half the sample rate is the highest frequency that can be recorded, which is why 44.1 kHz covers the full 20 kHz range of hearing. Try it: FFT & Sampling Calculator →
Span
The width of the frequency range shown across the screen, from left edge to right edge. A span of 100 MHz gives an overview of a band; a span of 100 kHz zooms in on one signal. Narrow spans usually go with a narrow RBW. Read: Spectrum Analyzer Settings →
Spectrogram
A picture of frequencies over time: time runs across, frequency runs up, and colour shows level. It shows when sounds start and stop, how a pitch slides and which noises come and go. It is also called a waterfall when it scrolls from top to bottom. Try it: Spectrogram →

T

THD
Total harmonic distortion: how much of a device's output is harmonics that were not in the input, as a percentage or in dB. Play a pure 1 kHz tone, measure the peaks at 2, 3, 4 kHz and up, and compare them with the 1 kHz peak. Try it: THD / Distortion Analyzer →
True peak
The highest level a signal reaches between its digital samples once it is turned back into a smooth wave. It can be higher than the largest sample, so a file that never touches 0 dBFS can still clip on playback. Streaming guidelines ask for at most −1 dBTP. Try it: LUFS Loudness Meter →

V

VBW
Video bandwidth: a smoothing filter applied after an analyzer detects each signal. A narrow VBW makes a noisy trace calmer and easier to read, without changing the true signal levels. It works much like averaging on an FFT analyzer. Read: Spectrum Analyzer Settings →
VSWR
Voltage standing wave ratio: how well an antenna or a load matches its cable. 1:1 is a perfect match, where all the power goes into the antenna. At 2:1, about 11 % of the power is reflected back. Most transmitters want 2:1 or better. Try it: VSWR Calculator →

W

White noise
Random noise with equal energy at every frequency, so it sounds bright and hissy. On an FFT spectrum it reads flat, while on an octave-band RTA it rises 3 dB per octave, because each band up holds twice as many Hz. Try it: Audio File Spectrum Analyzer →
Window function
A shape applied to each piece of signal before the FFT, fading its edges so the cut does not create false frequencies. Each window trades sharp peaks against low leakage: Hann is the common default, flat-top reads levels best, and Blackman keeps leakage lowest. Try it: Window Function Explorer →

Written and tested by the SpectrumAnalyzer.app team · Last checked