What is a sound spectrum?

A sound spectrum is a representation of the energy present at different frequencies in an audio signal. Instead of showing amplitude over time, it shows which frequencies are dominant — for example, a bass-heavy recording will show high energy in the 20–250 Hz range. Spectrum analysis helps diagnose microphone coloration, room acoustics, and signal noise.

What do the frequency bands (Bass, Mid, Treble) represent?

Bass covers 20–250 Hz (low-end rumble, kick drums, bass guitar), Mid covers 250 Hz–4 kHz (vocals, guitars, most musical content), and Treble covers 4–20 kHz (air, cymbals, high-frequency detail). A balanced microphone should capture energy across all three bands without excessive coloration in any one range.

What is A/C/Z frequency weighting?

Weighting curves adjust measured dB levels to account for how humans perceive loudness at different frequencies. A-weighting (dBA) mimics the human ear's reduced sensitivity at low and very high frequencies and is used for environmental noise measurements. C-weighting (dBC) is flatter and used for peak measurements. Z-weighting (dBFS) is completely flat with no adjustment applied.

What is the crest factor and why does it matter?

The crest factor is the difference between the peak level and the RMS (average) level of a signal, expressed in dB. A high crest factor means the signal has sharp transients relative to its average loudness. Pink noise has a crest factor around 12–15 dB, while a sine wave is about 3 dB. Knowing the crest factor helps set appropriate headroom when recording.

What is FFT size and how does it affect the analysis?

FFT (Fast Fourier Transform) size determines the frequency resolution of the spectrum analysis. A larger FFT size (e.g. 16384) provides finer frequency detail, allowing you to distinguish closely spaced tones, but requires more processing time. A smaller FFT size (e.g. 2048) responds faster to changes but shows coarser frequency resolution.

What is the difference between Pink Noise and White Noise for testing?

White noise has equal energy per frequency (flat spectrum), making it useful for testing system frequency response. Pink noise has equal energy per octave band, which sounds more balanced to human ears and is preferred for room acoustic measurements and speaker/microphone calibration. Use pink noise when measuring a room or microphone's natural response.

What is 1/3 octave resolution versus 1/1 octave?

Octave resolution determines how finely the frequency spectrum is divided into bands. 1/1 octave gives 10 broad bands across the audible range — useful for quick overviews. 1/3 octave gives 31 narrower bands, revealing more detail about where specific frequency problems exist. Room acoustic measurements and equalizer tuning typically use 1/3 octave analysis.

What does smoothing do to the spectrum display?

Smoothing averages the spectrum over time so the display is less jittery and easier to read. A smoothing value of 0 shows raw, unaveraged data that updates very quickly. Higher smoothing values (toward 0.95) produce a slow, stable trace that is easier to interpret but less responsive to sudden transient events. Use low smoothing to catch peaks, and high smoothing to assess steady-state levels.

Microphone Quality and Spectrum

The Microphone Quality and Spectrum analyzer shows your mic's real frequency response with your browser's noise suppression and echo cancellation turned off, not a smoothed-over version: click Start Microphone to watch the live spectrum, then Measure Noise Floor (2s) to capture a snapshot. You'll get your Dominant Frequency, Noise Floor, Sample Rate, and automatic detection of 50/60 Hz Mains Hum from a nearby power source. Check it before a recording session to catch a bad cable, ground loop hum, or an oddly-shaped frequency response. Use the record and play back your mic to check your hardware is actually working the way you expect, right from this page.

Microphone Quality & Spectrum

Live frequency spectrum, logarithmic axis (20 Hz – ~20 kHz)

20 Hz200 Hz2 kHz20 kHz

Click Start Microphone and allow access when your browser asks.

Readings

—Dominant Frequency

—Noise Floor

—Mains Hum

—Sample Rate

Stay silent for 2 seconds while measuring — this captures your room's background noise level for comparison.

Why Processing Is Disabled

This test deliberately requests your microphone with echo cancellation, noise suppression and automatic gain control all turned off. Those features are on by default in most calling apps, but they actively reshape the frequency spectrum — leaving them on here would measure your browser's processing, not your microphone.

Mains hum shows up as a spike at exactly 50 Hz (most of the world) or 60 Hz (North America and parts of Asia) — usually from a nearby power supply, unshielded cable, or ground loop.

Whether you're troubleshooting room acoustics in a home studio, checking a podcasting setup for unwanted hum, or performing a quick hearing check, understanding microphone quality and spectrum gives you an immediate, visual window into the input signal your device is capturing. This online audio spectrum analyzer turns raw mic input into a real-time frequency spectrum display — letting you see exactly what frequencies are present, how loud each band is, and what your environment's acoustic fingerprint looks like in real time.

Free Audio Spectrum Analyzer — Understand Microphone Quality and Spectrum from Your Browser

The browser-based sound analyzer works entirely in-browser, meaning all processing is performed locally with no server upload of your data: your recordings are never transmitted, never stored, and never saved. It uses the Web Audio API to apply a Fourier transform — specifically an FFT — to the incoming time-domain input, converting it into the spectral domain so you can see which tonal bands carry the most energy at any given moment. This is the same underlying physics principle used in professional acoustic tools, delivered through a real-time display that requires no software installation. Use the check microphone input gain to check your hardware is actually working the way you expect, right from this page.

Grant Microphone Access and Start Your Frequency Spectrum Analysis

To begin capturing input from your mic, click Start Analysis and choose Live Microphone as your source. Your browser will prompt you for microphone permission — click Allow when the dialog appears. This step is mandatory: without it the analyzer cannot receive the input from your device's mic. Once permission is granted, the live FFT visualization launches immediately and the spectral display begins plotting in real time. You can also select Upload Audio File if you prefer to examine an existing recording rather than a live mic feed — this is ideal for reviewing captured tracks or preset clips. The mic on/off toggle lets you pause and resume capture without losing your current settings, and the toggle button is clearly visible at the top of the control panel.

Choose your source: Select Start Mic Analysis for live capture or Upload Audio File for an existing sound file.
Grant permission: Click Allow when your browser asks for microphone access. Your privacy is fully protected — no recording is made, and data is not stored anywhere.
Read the display: Watch the real-time plot as it maps every band from 20 Hz up to 20 kHz. Bar height reflects loudness in dBFS.
Identify the dominant note: The analyzer pinpoints the highest-energy bin and labels it as a musical note — including whether the tone is cents sharp or cents flat from the ideal tuning reference (A4 440 Hz).

A disclaimer is displayed beneath the tool: the analyzer provides visualization and relative measurement, not a substitute for a calibrated SPL meter conforming to IEC 61672. A and C weighting curves are computed via biquad approximation, which is accurate enough for spatial acoustics diagnostics and environmental survey work but should not be cited as calibrated SPL data in formal noise standards compliance reports.

Reading the Sound Capture Across Low, Mid, and High Bands

The tonal range is divided into three primary bands, each covering a distinct perceptual range that corresponds to how our ears process sound. Understanding what each band contains helps you diagnose problems quickly during mic analysis:

Low (20–250 Hz): This is the territory of bass frequencies, sub-bass, and low content — percussion, room rumble, low piano notes, and engine noise live here. A spike at 50 Hz hum or 60 Hz typically signals electrical interference from poor shielding or a ground loop. Room modes and tonal response issues almost always appear in this region, and can risk speaker damage if driven hard.
Mid (250–4,000 Hz): The mid range covers most human speech, musical instruments, and the bands critical for speech clarity. Problems around 300 Hz and 500 Hz often produce a boxy mix; the 4 kHz region strongly influences presence and intelligibility. This range is the backbone of vocal analysis and broadcasting quality checks.
High (4,000–20,000 Hz): High frequencies, treble, and upper overtones — cymbals, sibilance, and the airy upper partials that give captures their sparkle. The 250 Hz crossover boundary between low-mid and upper-mid is visible in the display when you switch to octave bands view.

Every tone produces a unique combination of these ranges — a kind of tonal fingerprint or sound fingerprint — visible as the live plot scrolls. You can switch between a logarithmic scale and a linear scale using the checkbox in the View panel; logarithmic is generally preferred because it matches how our ears perceive tone and octave doubling.

View Settings, Stats, and Display Options

The View & Settings panel gives you precise control over the FFT analyzer's behavior. Adjust the FFT size to trade resolution against responsiveness — a larger FFT size yields finer bins and higher spectral resolution, at the cost of a slightly slower update rate. The smoothing slider controls how aggressively the decay rate is applied between frames, producing either a snappy real-time response or a smoother averaged trace. The sample rate is determined by your hardware and browser, and the display automatically adapts its scale accordingly.

Signal Stats sit below the display and update continuously during live capture:

Peak: The absolute highest sample level, shown in dBFS. Useful for detecting clipping or peak measurement during loud transients.
RMS: Root Mean Square — a measure of average power that reflects perceived loudness far better than peak alone. RMS measurement is the foundation of loudness normalization in mastering and broadcast workflows.
Crest Factor: Peak minus RMS, expressed in dB. High crest factor values indicate dynamic content with loud transients; low values suggest heavy compression or a very steady source like broadband noise.
Correlation: Measures phase relationship between left and right channels — useful for spotting phase cancellation.
Clipping indicator: Flags when the input exceeds 0 dBFS, which causes audible distortion in the captured output.

Switch to Spectrogram view to see a scrolling time-versus-pitch plot: the horizontal axis becomes time, the vertical axis shows tonal bands, and colour intensity maps to amplitude — high intensities appear bright orange or yellow, while low intensities show as dark blue or black. The spectrogram is invaluable for spotting voice patterns, sporadic interference, and transient events that the live FFT display may not clearly reveal. This visualization technique is used in professional audio engineering and sound design studios.

Freeze, Save, and Compare Measurements

The freeze visualization button pauses the real-time display so you can inspect a specific moment without the graph continuing to update — ideal for detailed spectral analysis of a transient event. To build a comparison, click Save Ref while measuring your first scenario: this captures a reference trace (displayed in cyan) that overlays the live plot. After making changes — repositioning the mic, adjusting EQ, swapping a speaker — click Save Ref again to capture a second reference trace (magenta). Viewing both against the live output lets you immediately assess before/after EQ differences or validate a placement improvement. A third save rolls the oldest reference off automatically. This save reference workflow is also useful as an EQ comparison tool when optimizing room treatment or speaker placement.

All captured data is available via the Export menu as CSV exports. Each export includes the current spectral snapshot, octave band data, the Leq session log, and any saved references. Header lines embed the measurement timestamp, sample rate, FFT size, and weighting, making every export fully reproducible in external tools such as Excel, Google Sheets, Python, or R. A PNG screenshot of the current display is also available from the same menu for documentation or reporting purposes.

Advanced Measurement Tools and Test Signal Features for Frequency Spectrum Assessment

Test Signal Generator: Pink Noise, Broadband Noise, Sine Wave, and Sweep

The built-in test signal generator emits four waveform types through your speakers or headphones, allowing you to perform speaker testing, spatial acoustic checks, level calibration, and speaker measurement directly within the browser. Always lower your speaker volume before pressing play — high-level test tones can cause speaker damage and risk auditory harm, particularly with a logarithmic sweep at full level.

Pink noise: Has equal energy per octave, making it the industry-standard waveform for room tonal response and pink noise measurement. Pair it with mic capture to visualize your room's complete response curve. Because it has equal energy per octave rather than equal energy per Hz, pink noise sounds balanced rather than harsh.
White noise: Carries equal energy per Hz across the full audible range, making it brighter and more appropriate for system stress testing and electronics evaluation.
Sine wave: A pure sine wave at any chosen band — a tone is ideal for single-point level calibration, perceptual checks, instrument tuning, or driving a speaker at one specific note to check for room modes. You can also use a tone generator in a companion tab and monitor its output here.
Log sweep: Sweeps continuously from the lowest to the highest audible range over ten seconds. Watch the live display for resonances, rolloff, or non-linearity — this is the most revealing test tone for diagnosing speaker coloration and spatial acoustic problems. A sweep can also expose overtone content generated by a room, which shows up as energy at higher bands than the fundamental being swept.

Worked example — home studio room acoustics: A producer grants mic access, clicks Start Analysis, and immediately notices a pronounced resonance in the low band near 80 Hz on the live spectrogram. They switch the test generator to pink noise, lower speaker volume, press Play, and watch the tonal map fill out. A clear standing wave peak at 80 Hz confirms a room mode — the vibration bouncing between two parallel walls. They click Save Ref, add acoustic panels, then capture a second reference: the overlay confirms the peak has dropped by 6 dB, validating the treatment without any external hardware.

Noise Measurement Session: Leq, Percentile Levels, and Statistical Descriptors Explained

Activating the Noise Measurement Session starts a timed logging mode that builds a statistical picture of your environment's levels — essential for environmental monitoring, occupational health assessments, and noise pollution documentation. The session calculates several percentile descriptors:

Leq (Energy Equivalent Level): The time-averaged level over the full measurement duration. This is the single most important number for assessing the overall dose of noise exposure, used in workplace noise regulations and environmental survey standards worldwide.
L10: The level exceeded 10% of the measurement time — representing the loud events and peaks. Relevant for assessing traffic noise impact and building noise complaints.
L50: The median level — exceeded exactly half the time. A useful midpoint reference for characterizing noise variability.
L90: The level exceeded 90% of the time — effectively the ambient floor of the environment. A quiet space with a low ambient floor is ideal for podcasting, broadcasting, or clinical environments used by patients and doctors.
Lmax and Lmin: The absolute highest and lowest samples captured, useful for spotting impact noise events or sporadic disturbances during the session.

For steady sources like HVAC noise, 30 seconds is typically sufficient. For variable conditions — office environments, restaurant settings — five minutes gives a more representative average. Formal survey standards for environmental monitoring typically require 15 minutes of continuous logging. The gap between your upper and lower percentile values tells you how variable the noise environment is: a wide gap indicates highly sporadic interference; a narrow gap means a steady source.

Worked example — podcaster assessing ambient noise: A podcaster wants to assess the background noise floor of their capture space before a session. They record system output from a YouTube video by clicking Share Tab, ticking "share audio" in the browser dialog, and selecting the correct tab. After a 30-second session, the Leq reads −42 dBFS and the ambient floor shows −51 dBFS — confirming a clean, low noise floor suitable for capture. The session log is exported as a CSV and attached to the project file for future reference.

Mid/Side Stereo Analysis for Microphone Quality Assessment

Switching to the Mid/Side mode reveals a powerful dimension of microphone quality and spectrum that a standard stereo display hides. The mid channel contains everything identical in the left and right channels — essentially the mono content of the input, including centred vocals, kick drums, and dialogue. The side channel contains the stereo difference — the elements that create stereo width and spatial depth.

A weak side output means the capture is nearly mono compatible — this is typical of single-mic podcasting setups and is generally desirable. A strong side channel relative to mid indicates wide stereo content, which can cause phase problems on mono playback systems. In music production and mixing, monitoring the mid/side balance helps engineers ensure tracks translate correctly across mono and stereo playback. For mid/side analysis in a home studio context, a correlation value near +1 confirms strong mono compatibility; values approaching −1 indicate potential phase cancellation.

A/C/Z Weighting and Octave Bands for Sound Spectrum Analyzer Online Work

The weighting selector applies a standardized response curve to the incoming input before level calculations are performed. Each weighting curve serves a distinct purpose in acoustic measurement and psychoacoustics:

A-weighting (A): Models the sensitivity of our ears at moderate listening levels. It attenuates bass and very high bands, emphasizing the range where perception is most sensitive (roughly 1–4 kHz). A-weighting is the standard for workplace noise regulations, environmental monitoring, and occupational health assessments. An online audiometer used by doctors for a home auditory check typically applies A-weighting curves for threshold measurements.
C-weighting (C): Much flatter at low bands than A, capturing bass content more accurately. C-weighting is used for peak measurement and impact noise assessment where bass energy matters, such as construction noise or low-band HVAC analysis.
Z-weighting (Z): Completely flat — raw dBFS with every band treated equally. Z-weighting (formerly called "linear") is useful for pure spectral analysis and situations where you want uncoloured data without psychoacoustic correction.

The octave bands display groups the full audible range into either 1/1 octave (ten broad bands) or 1/3 octave (31 finer bands). Because each octave represents a doubling of pitch, octave-band grouping matches perceptual relationships. The 1/3 octave resolution is widely used in spatial acoustics, building noise, and architectural work because it provides enough detail to diagnose problems without the noise of full FFT resolution. You can export octave band data as a CSV file for use in acoustic modelling software or compliance reporting against noise standards.

Worked example — speaker measurement with pink noise: An audio engineer runs the test signal generator set to pink noise, ensures the volume is reduced first as warned, then clicks Play. The analyzer captures the room's tonal response via the live mic. Switching to 1/3 octave bands reveals a 4 dB dip around 200 Hz and elevated energy at 4 kHz — classic signs of a poorly positioned monitor and early reflections. They export the octave band CSV, open it in a spreadsheet, and plot the before/after correction curve after repositioning the speaker. The FFT engine uses parabolic interpolation — the same sub-bin interpolation technique — to refine dominant pitch and note estimates beyond individual bins, which is why note detection remains accurate even at low bass fundamentals where bins are widely spaced.

The harmonic product spectrum (HPS) algorithm underpins accurate note detection: rather than simply picking the highest-energy bin, HPS multiplies the spectral output with downsampled versions of itself to reinforce the fundamental while suppressing isolated overtones. This means that a violin or whistling performance with rich overtone content — many partials above the fundamental — still resolves to the correct musical note rather than a higher overtone. A pure sine wave produces a single spike with no overtone content, while complex sources like percussion, piano notes, or bird songs reveal a dense comb of partials across the tonal domain. Even sounds as distinctive as whale clicks, dolphin clicks, police sirens, or a dial-up modem each produce a unique and recognisable tonal fingerprint in the spectrogram.

Whether you are a professional working in sound design, mastering, or audio engineering, a student exploring the physics of music and tone for education purposes, or simply curious about the input captured by your device, this spectrum analyzer online provides an engaging platform for exploring spectral analysis without any installation. The tool is fully browser-based, supporting real-time analysis of live mic input, system output via tab capture, or any uploaded file you choose to load directly into the player. Use the play button and pause button to control playback of uploaded clips or preset files, and watch the tonal patterns, waveform, and displacement unfold in the visualization panel. This FFT analyzer, tonal analyzer, and acoustic analyzer in one package reflects the breadth of modern signal processing available directly in your browser — a true spectral visualizer that empowers anyone to analyze sound, plot tonal data, and develop genuine analytical skills through hands-on exploration.

Frequently Asked Questions

What is a sound spectrum?: A sound spectrum is a representation of the energy present at different frequencies in an audio signal. Instead of showing amplitude over time, it shows which frequencies are dominant — for example, a bass-heavy recording will show high energy in the 20–250 Hz range. Spectrum analysis helps diagnose microphone coloration, room acoustics, and signal noise.
What do the frequency bands (Bass, Mid, Treble) represent?: Bass covers 20–250 Hz (low-end rumble, kick drums, bass guitar), Mid covers 250 Hz–4 kHz (vocals, guitars, most musical content), and Treble covers 4–20 kHz (air, cymbals, high-frequency detail). A balanced microphone should capture energy across all three bands without excessive coloration in any one range.
What is A/C/Z frequency weighting?: Weighting curves adjust measured dB levels to account for how humans perceive loudness at different frequencies. A-weighting (dBA) mimics the human ear's reduced sensitivity at low and very high frequencies and is used for environmental noise measurements. C-weighting (dBC) is flatter and used for peak measurements. Z-weighting (dBFS) is completely flat with no adjustment applied.
What is the crest factor and why does it matter?: The crest factor is the difference between the peak level and the RMS (average) level of a signal, expressed in dB. A high crest factor means the signal has sharp transients relative to its average loudness. Pink noise has a crest factor around 12–15 dB, while a sine wave is about 3 dB. Knowing the crest factor helps set appropriate headroom when recording.
What is FFT size and how does it affect the analysis?: FFT (Fast Fourier Transform) size determines the frequency resolution of the spectrum analysis. A larger FFT size (e.g. 16384) provides finer frequency detail, allowing you to distinguish closely spaced tones, but requires more processing time. A smaller FFT size (e.g. 2048) responds faster to changes but shows coarser frequency resolution.
What is the difference between Pink Noise and White Noise for testing?: White noise has equal energy per frequency (flat spectrum), making it useful for testing system frequency response. Pink noise has equal energy per octave band, which sounds more balanced to human ears and is preferred for room acoustic measurements and speaker/microphone calibration. Use pink noise when measuring a room or microphone's natural response.
What is 1/3 octave resolution versus 1/1 octave?: Octave resolution determines how finely the frequency spectrum is divided into bands. 1/1 octave gives 10 broad bands across the audible range — useful for quick overviews. 1/3 octave gives 31 narrower bands, revealing more detail about where specific frequency problems exist. Room acoustic measurements and equalizer tuning typically use 1/3 octave analysis.
What does smoothing do to the spectrum display?: Smoothing averages the spectrum over time so the display is less jittery and easier to read. A smoothing value of 0 shows raw, unaveraged data that updates very quickly. Higher smoothing values (toward 0.95) produce a slow, stable trace that is easier to interpret but less responsive to sudden transient events. Use low smoothing to catch peaks, and high smoothing to assess steady-state levels.