Return to samples with ifft

ifft(x) computes the inverse fast Fourier transform. It converts complex frequency coefficients back into a time- or index-domain sequence, reversing the operation performed by fft.

Round-trip example

Transform a sample sequence with fft, then pass the result to ifft. The reconstructed values should match the original apart from small floating-point imaginary residues or rounding differences. Use re when a theoretically real reconstruction displays negligible imaginary noise, and abs to inspect spectral magnitudes before inversion.

Conventions matter

IFFT expects coefficients in the matching ordering and normalization convention. Do not edit, sort, or truncate bins without understanding the resulting signal change. Frequency filtering is commonly performed by transforming, modifying selected coefficients, then inverting; freqz instead analyzes a filter’s designed response.

Input caveats

The input should be a compatible numeric/complex sequence. Result length and sample-rate interpretation come from the original transform setup, not from ifft alone. Preserve dimensions and avoid assuming that a real-looking spectrum guarantees a real output unless conjugate symmetry is present.