What does zero-padding actually do to the FFT?

Zero-padding appends zeros to the time-domain signal before the FFT. This increases the number of frequency bins in the output, which interpolates between the original DFT frequencies. It does not improve the true resolution (which depends on the signal length), but it interpolates between DFT bins to make the peak center easier to locate accurately without altering the underlying peak width.

Why must the FFT length be a power of 2?

The Cooley-Tukey FFT algorithm requires the transform size to be a power of 2 for maximum efficiency. Using a non-power-of-2 size forces a slower mixed-radix algorithm or requires manual padding. Rounding up to the next power of 2 is nearly always the right choice.

How much zero-padding is enough?

A 2x zero-pad (doubling to the next power of 2) is the most common choice and gives a 2x interpolation factor. A 4x pad doubles that again. Beyond 4x the returns diminish rapidly — you pay more computation for marginal improvement in peak readability.

Does zero-padding improve frequency resolution?

No. True frequency resolution — the minimum separation between two distinguishable spectral peaks — is set by the signal length (1/T). Zero-padding only interpolates between fixed DFT frequencies; it cannot reveal structure that the signal length does not support.

What is the interpolation factor?

The interpolation factor is the ratio of the padded FFT size to the unpadded FFT size (next power of 2 at or above the signal length). A factor of 2 means twice as many bins span the same frequency range, so spectral peaks are sampled at twice the density.

Can I use a length that is not a power of 2?

You can, but most FFT libraries are significantly slower on non-power-of-2 sizes. This tool always recommends power-of-2 padded lengths for compatibility and performance. If your library explicitly supports prime-factor FFTs (e.g. FFTW), a non-power-of-2 size may still be fast, but the rule-of-thumb still favors powers of 2.

Zero-Padding Helper

Enter your signal length and get the recommended zero-padded FFT sizes with spectral interpolation ratios and frequency resolution.

Signal Parameters

Signal Length (samples) *

Number of samples in your input signal (N).

Sample Rate (Hz) — optional

Leave blank to show normalized (cycles/sample) resolution.

Common Signal Lengths

Enter a signal length and click Calculate.

Copied!

Summary

Zero-padding a signal before an FFT increases the number of output bins without adding new spectral information — it interpolates between the true DFT frequencies, making peaks easier to locate and read. This tool takes your raw signal length, computes the nearest power-of-2 FFT sizes (unpadded, 2x, and 4x padding), and shows the resulting bin count, frequency resolution, and interpolation factor for each option. Enter your sample rate to see resolution in Hz, or leave it blank for normalized values.

How it works

Enter the number of samples in your signal (its length N).
Optionally enter the sample rate in Hz to display frequency resolution in Hz.
Click Calculate. The tool finds the next power of 2 at or above N (the minimum FFT size).
It also computes 2x and 4x padded lengths (the next two higher powers of 2).
For each padded length it shows bin count, Δf (or normalized bin spacing), and the interpolation factor vs the unpadded case.
Pick the padded size that balances resolution improvement with acceptable computation cost.

Use cases

Choose an optimal power-of-2 FFT size to ensure high computational efficiency for non-power-of-2 signal lengths.
Improve peak frequency estimation accuracy in audio analysis by interpolating between bins.
Select zero-padding ratio for a vibration analysis STFT where bin resolution matters.
Quickly find the smallest power-of-2 FFT size that fits a given block length.
Compare unpadded vs padded frequency resolution before running a batch spectrum job.
Verify zero-padding ratios meet a minimum interpolation factor for a project spec.

Frequently Asked Questions

Last updated: 2026-06-19 · Reviewed by Nham Vu