ShiftTools ShiftTools
Zero-Upload Browser Utility

Audio Spectral Analyzer

Analyze audio frequencies in real time with a 4-panel FFT spectrum, 2D waterfall spectrogram, oscilloscope, stereo phase scope, and pitch detector.

Interactive Tool Workspace

Loading local in-browser WebAssembly engine & hardware cryptographic sandbox...

Drop an audio file, enable your live microphone feed, or trigger the built-in 18-mode signal synthesizer to run a real-time audio spectrum analyzer directly in your browser. Four synchronized WebGL/Canvas instruments - an FFT Frequency Spectrum with harmonic markers, a 2D Waterfall Spectrogram, a Time-Domain Oscilloscope, and a Stereo Lissajous Phase Scope - update at 60 frames per second alongside peak dBFS meters, RMS loudness, and fundamental pitch detection.

When visual frequency analysis catches what studio headphones miss

Human ears adapt quickly to room resonances and low-end rumble, which makes it easy to miss a 50 Hz or 60 Hz AC power hum, a sub-bass buildup at 35 Hz, or a high-frequency 15.7 kHz coil whine sitting inside a voice recording. Looking at a high-resolution Fast Fourier Transform (FFT) curve and a scrolling 2D waterfall spectrogram exposes continuous tonal lines and transient clicks immediately, even when you are editing on laptop speakers.

Mixing engineers and podcasters also run into a silent trap when collapsing stereo tracks to mono on phone speakers: out-of-phase stereo cancellation. If a stereo microphone pair or widening plugin flips the polarity between the Left and Right channels, the track sounds wide in headphones but vanishes or turns hollow on a mono Bluetooth speaker. Watching the Stereo Vectorscope (Lissajous Goniometer) and the -1.0 to +1.0 phase correlation readout tells you within two seconds whether your stereo image is mono-compatible.

And when you do not want to upload a file at all - such as testing room acoustics, tuning a musical instrument against 440 Hz Concert A, or checking where your headphones roll off in the sub-bass - switching to the Live Microphone or Synthetic Generators tab turns your browser into a self-contained acoustic test bench.

How to use the audio spectrum analyzer step by step

Method 1: Inspect an Audio Recording & Freeze Transient Peaks (Audio File Tab)

  1. Load your recording: Select the Audio File source tab at the top-left of the rack and drag in an MP3, WAV, FLAC, M4A, OGG, AAC, or WebM file (up to 500 MB).
  2. Play or jump to a timestamp: Press Play on the transport bar, or click the clock icon to jump to an exact mm:ss.ms frame where a hum, click, or sibilant peak occurs.
  3. Inspect frequencies and harmonics: Toggle the FFT Spectrum between Logarithmic and Linear scales, turn on the Harmonics (f0..5f0) overlay to trace overtones, and watch the Stereo Phase Scope (-1.0 to +1.0) for mono compatibility.
  4. Freeze and measure: Click Freeze Spectrum at any moment to lock all four 60 FPS visualizers in place, then hover your mouse crosshair over any spike on the FFT graph to read its exact frequency (Hz), musical note, and amplitude (dBFS).

Method 2: Measure Room Acoustics or Tune Instruments (Live Microphone Tab)

  1. Enable the microphone input: Switch to the Live Microphone tab and grant browser permission for your audio interface or built-in mic (wear headphones or lower speaker volume to prevent feedback loops).
  2. Track fundamental pitch (f0) in real time: Play a note or speak into the mic while watching the Autocorrelation Pitch Detector readout for the detected musical note (e.g., A4 · 440.0 Hz) and the tuning deviation needle (±50 cents).
  3. Spot continuous room modes and electrical hum: Watch the scrolling 2D Waterfall Spectrogram—persistent horizontal thermal lines reveal constant 50 Hz / 60 Hz AC hum, HVAC rumble, or standing room resonances.

Method 3: Run Acoustic Reference Signals & Live Biquad DSP Filters (Synthetic Generators Tab)

  1. Select one of the 18 built-in test generators: Open the Synthetic Generators tab and choose a reference waveform (1 kHz Sine, 440 Hz A4, 60 Hz Sub-Bass, Square, Sawtooth), an acoustic noise color (Pink Noise -3dB/oct, White, Brown), or a diagnostic sweep (20Hz–20kHz Log Sweep, SMPTE IMD, 360° Stereo Lissajous).
  2. Engage a real-time Biquad DSP Filter: Click any of the 8 preset filters in the Preset DSP Filters rack (Sub Low-Pass 120Hz, Rumble Cut HP 100Hz, Telephone Band 300–3.4kHz, 60Hz Hum Notch) and drag the Cutoff (fc) and Resonance (Q) sliders to hear and see the filter slope reshape the spectrum live.
  3. Adjust FFT resolution: In the bottom-right Configuration card, increase FFT Window Size up to 16384 samples (2.93 Hz bin precision) for surgical bass analysis or drop to 512 samples for fast drum transients.

Four synchronized visualizers, 18 test generators, and live DSP filters

Rather than showing a decorative equalizer animation, every panel in this workspace is wired directly to the Web Audio API's AnalyserNode, BiquadFilterNode, and stereo channel splitter.

  • FFT Frequency Spectrum with Harmonic Series overlay: Plots real-time decibel magnitude from 20 Hz to 20 kHz on a logarithmic or linear horizontal axis. Enabling the Harmonics badge automatically locks onto the fundamental frequency ($f_0$) and highlights its 2nd, 3rd, 4th, and 5th overtones ($2f_0$ to $5f_0$).
  • 2D Time-Frequency Waterfall Spectrogram: Scrolls frequency vertically against time horizontally, mapping decibel intensity to a thermal color gradient (deep navy for silence through cyan, amber, and white-hot peaks) to reveal plosives, sibilance, and intermittent noise.
  • Time-Domain Oscilloscope & Stereo Lissajous Vectorscope: The oscilloscope plots instantaneous waveform voltage ($[-1.0, +1.0]$) to spot hard digital clipping, while the XY Lissajous Goniometer plots Left versus Right channel amplitude at 45° axes to visualize stereo width and phase coherence.
  • 18 built-in Synthetic Signal Generators: Generate reference waveforms (1 kHz Sine, 440 Hz A4, 60 Hz Sub-Bass, Square, Sawtooth, Triangle), six acoustic noise profiles (White, Pink -3dB/oct, Brown -6dB/oct, Blue +3dB/oct, Violet +6dB/oct, Velvet Impulse), and complex test signals (20Hz–20kHz Log Sweep, SMPTE IMD 60Hz+7kHz, Octave Formant, FM Sideband, Harmonic Major Triad, and 360° Stereo Lissajous).
  • Autocorrelation Pitch Detector & True Peak Telemetry: Displays the detected musical note (e.g., A4, C#3), exact frequency in Hz, tuning deviation needle in cents ($pm 50$ cents), Left/Right Peak dBFS bars, RMS loudness, and Spectral Centroid brightness.

Worked example: tracking down a 120 Hz hum and stereo phase issue

Suppose you drop a 45-second interview recording named Studio-Interview-Raw.wav (48 kHz, 16-bit Stereo, 8.2 MB) into the Audio File dropzone and press Play:

  1. Looking at the FFT Frequency Spectrum in Logarithmic view during a quiet pause between sentences, you notice a sharp vertical spike sitting consistently at 120 Hz peaking at -28.4 dBFS, with a second harmonic spike at 240 Hz - a classic full-wave rectified 60 Hz power supply hum.
  2. Hovering your mouse over the spike reads out 120 Hz · B2 · -28.4 dB in the top-right badge. You click Freeze Spectrum to lock the frame and inspect the exact harmonic spacing.
  3. Unfreezing playback, you click the Rumble Cut HP (100Hz) preset in the Preset DSP Filters bar and slide the cutoff frequency $f_c$ to 135 Hz ($Q = 1.2$). Watching the FFT spectrum live, the 120 Hz spike drops below -64 dBFS without thinning out the speaker's fundamental vocal range around 165 Hz.
  4. Glancing at the Stereo Phase Scope, you notice the Lissajous cloud is tilted horizontally with a phase readout of -0.38, indicating one microphone cable was wired out of phase before you export the track to the Audio Converter.

Choosing FFT window sizes and understanding browser audio limits

In digital signal processing, choosing the FFT Size (adjustable in the bottom-right Configuration card from 512 up to 16384) always involves a trade-off between frequency resolution and time resolution:

  • At an audio sample rate of $48,000\text{ Hz}$, an FFT size of 2048 divides the spectrum into 1024 frequency bins spaced $48000 / 2048 = 23.44\text{ Hz}$ apart, updating roughly every 42 milliseconds. That is fast and responsive for general music and speech.
  • Increasing the FFT size to 16384 narrows each bin down to $2.93\text{ Hz}$, allowing you to distinguish two closely spaced bass notes at 55 Hz and 58 Hz - at the cost of a slightly slower 341-millisecond time window.
  • Dropping the FFT size to 512 gives ultra-fast 10.6-millisecond transient response for drum hits, with wider 93.75 Hz frequency bins.

Also note that when using the Live Microphone tab, browsers apply default OS echo-cancellation or automatic gain control unless disabled; keep your speaker volume low or wear headphones when testing live mic feeds so you do not create an acoustic feedback loop.

Complete local audio processing

Whether you load a 400 MB studio WAV file or grant microphone access, 100% of the decoding, Fast Fourier Transform math, biquad filtering, and canvas rendering happen locally inside your browser's Web Audio API context. Zero audio frames or voice recordings are ever transmitted over the network.

Related audio, media, and RF tools

Once you inspect your audio track, you can normalize loudness, trim silence, or transcode between MP3, WAV, FLAC, and OGG in the Audio Converter, extract audio tracks from video files in the Video Converter, or shrink heavy recordings in the Video Compressor & Trimmer. For RF spectrum heterodyne calculations, check the Satellite LNB Calculator, or read our engineering deep dive on how to analyze audio frequencies and spectrum online.

Frequently Asked Questions

What is the difference between the Logarithmic and Linear scale buttons on the FFT Spectrum?

Human hearing perceives pitch logarithmically by octaves, where the distance from 100 Hz to 200 Hz sounds the same as 1,000 Hz to 2,000 Hz. Logarithmic mode spreads bass, midrange, and treble evenly across the graph, while Linear mode spaces every 1,000 Hz equally to inspect high-frequency harmonic spacing.

How do I read the Stereo Phase Scope (Lissajous Goniometer)?

A vertical line down the center (+1.0 correlation) means Left and Right channels are identical mono. A balanced circular cloud (+0.3 to +0.8) indicates healthy stereo width, while a horizontal flat cloud (negative correlation toward -1.0) warns of out-of-phase cancellation.

What is the difference between White, Pink, and Brown noise in the Synthetic Generators tab?

White noise has equal energy per Hertz (flat on a linear FFT, sounding bright to human ears). Pink noise rolls off at -3 dB per octave so every octave carries equal energy, making it the standard reference for studio monitor calibration. Brown noise rolls off at -6 dB per octave for a deep low-end rumble.

What does the Spectral Centroid readout in the Acoustic Telemetry card measure?

Spectral Centroid calculates the center of gravity of the frequency spectrum in Hertz. A dark, bass-heavy track will show a centroid around 400–800 Hz, whereas a bright vocal or cymbal-heavy mix shifts the centroid upward to 2,500–5,000 Hz.

Can I jump to an exact timestamp inside a long audio recording?

Yes. Click the clock icon next to the time readout in the transport bar to open the Jump to Time modal. You can type a timestamp in mm:ss.ms format (such as 02:14.5) or raw seconds (such as 134.5) and jump directly to that frame.

Does freezing the spectrum stop the audio so I can inspect a peak with my mouse?

Clicking Freeze Spectrum pauses both audio playback and all four visualizer canvases on the exact current buffer frame. While frozen, you can move your mouse cursor across the FFT graph to read the exact Hz, musical note, and dB level of any peak.