Published by PitchChanger · Updated · How these guides are made
How to use this workflow
Identify both numbers
Use sample rate in hertz and the number of samples in the transform. Do not use the song duration as FFT length.
Calculate the grid
Divide Fs by N. Bin k has nominal frequency k × Fs/N for the usual unshifted spectrum.
Check the window
Record the number of actual analyzed samples and the window function before interpreting nearby peaks.
A 48 kHz frequency grid at two transform lengths
With 4,096 samples, Δf = 48,000 / 4,096 = 11.71875 Hz and the data duration is 85.333 ms. With 16,384 actual samples the spacing becomes 2.9296875 Hz and the duration becomes 341.333 ms. Longer observation trades timing localization for finer spectral analysis.
A 440 Hz tone falls between bins on both grids: its nominal bin position is 37.5467 or 150.1867 respectively. A peak at a nearby bin is not sufficient evidence that the instrument is out of tune. This is a calculated illustration, not an exported spectrum.
Swipe or scroll sideways to read all columns.
| Fs / N | Bin spacing | Actual data window if unpadded |
|---|---|---|
| 48,000 / 4,096 | 11.718750 Hz | 85.333 ms |
| 48,000 / 8,192 | 5.859375 Hz | 170.667 ms |
| 48,000 / 16,384 | 2.929688 Hz | 341.333 ms |
Zero padding cannot add missing observation time
Padding 4,096 measured samples with zeros to a 16,384-point FFT produces a denser grid, but the observed sound is still only 85.333 ms long. Window shape, leakage, signal-to-noise ratio and observation duration affect separation of nearby components.
PitchChanger does not expose an FFT-length spectrum editor. Guitar Tuner provides a live single-note tuning workflow, not a bin inspector. Use an external spectrum analyzer to investigate bin spacing, and use the tuner for its supported note-and-cents task.