spectrogram — Compute a short-time Fourier transform spectrogram.

spectrogram(x) computes a short-time Fourier transform (STFT) of a signal. Optional arguments select the segment window, overlap, DFT length or explicit frequency grid, sampling frequency, frequency range, and PSD or power-spectrum scaling.

Syntax

s = spectrogram(x, win, nOverlap, freqSpec, Fs, freqRange, spectrumType)
[s, f, t, ps, fc, tc] = spectrogram(x, win, nOverlap, freqSpec, Fs, freqRange, spectrumType)

Inputs

Returns

Returned values from spectrogram depend on how many outputs the caller requests.

Errors

How spectrogram works

• s = spectrogram(x) returns the complex STFT matrix with time increasing across columns and frequency increasing down rows.
• [s, f, t, ps] = spectrogram(...) also returns the frequency vector, segment midpoint vector, and PSD or power-spectrum matrix.
• win may be a positive scalar length, a finite real window vector, or []. [] uses a Hamming window sized to divide the signal into about eight 50%-overlapped sections.
• nOverlap defaults to half the window length and must be less than the window length.
• freqSpec defaults to max(256, 2^nextpow2(length(win))). A real frequency vector with at least two elements may be supplied to evaluate explicit frequencies.
• Omitting Fs uses normalized radians/sample and segment midpoint times divided by 2*pi. Passing Fs as [] uses a 1 Hz sample rate.
• Real inputs default to one-sided output; complex inputs default to two-sided output and reject 'onesided'.
• The optional range selector accepts 'onesided', 'twosided', or 'centered'. Range selectors are not combined with explicit frequency vectors.
• The optional spectrum type accepts 'psd' (default) or 'power'; 'power' uses coherent-gain normalization.
• 'yaxis' is accepted as a plotting-orientation no-op for numeric execution. 'reassigned' is rejected until true energy-center reassignment is implemented.
• The documented signal storage class is single or double. Typed integer signals, windows, overlap counts, frequency specifications, and sample rates are role-specific RunMat extensions. Any value entering binary64 rejects if it would round.

Examples

STFT of a sine wave

fs = 1000;
t = (0:999) / fs;
x = sin(2*pi*100*t);
[s, f, tt, ps] = spectrogram(x, hamming(128), 64, 256, fs);

Use scalar window length

[s, f, t] = spectrogram(randn(1,1024), 128, 64, 256);

Centered complex STFT

x = exp(1i*2*pi*(0:127)/16);
[s, f, t] = spectrogram(x, 32, 16, 64, 'centered');

Power spectrum scaling

[s, f, t, p] = spectrogram(randn(1,512), 64, 32, 128, 'power');

Explicit normalized frequencies

w = linspace(0, pi, 64);
[s, wout, t] = spectrogram(randn(1,512), hamming(128), 64, w);

Using spectrogram with coding agents

Open a RunMat example with live inputs, then ask the agent to explain how spectrogram changes the result.

FAQ

Related Math functions

Open-source implementation

Unlike proprietary runtimes, every RunMat function is open-source. Read exactly how spectrogram is executed, line by line, in Rust.

• View the source for spectrogram in Rust on GitHub
• Learn how the RunMat runtime works
• Found a bug? Open an issue with a minimal reproduction.

About RunMat

RunMat is an open-source runtime that executes MATLAB-syntax code blazing on any GPU. It is licensed under the Apache 2.0 license.

• RunMat automatically optimizes your math for GPU execution on Apple, Nvidia, and AMD hardware. No code changes needed. Simulations that took hours now take minutes.
• Start running code in seconds. RunMat runs in the browser, on the desktop, or from the CLI. No license server, no IT ticket.

Getting started · Benchmarks · Pricing