fft-napi

Native FFT/IFFT for Node.js using Node-API (N-API).

Install

npm install

Usage

var fft = require('fft-napi').fft;
var ifft = require('fft-napi').ifft;

// Input is 1+0i 2+0i 3+0i 4+0i
var input = new Float32Array([1, 0, 2, 0, 3, 0, 4, 0]);
var spectrum = new Float32Array(input.length);
var restored = new Float32Array(input.length);

fft(input, spectrum, function (err, result) {
  if (err) throw err;
  // result === spectrum

  ifft(spectrum, restored, function (err2, restoredResult) {
    if (err2) throw err2;
    // restoredResult === restored
  });
});

API

fft(input, output, callback)

Forward complex FFT.

ifft(input, output, callback)

Inverse complex FFT (normalized by 1/N).

ifft applies the inverse transform and scales by 1/N internally, so if your input to ifft came from fft, then ifft(fft(x)) reconstructs x (within floating-point tolerance).

Example:

var fft = require('fft-napi').fft;
var ifft = require('fft-napi').ifft;

var time = new Float32Array([1, 0, 2, 0, 3, 0, 4, 0]);
var freq = new Float32Array(time.length);
var restored = new Float32Array(time.length);

fft(time, freq, function (err) {
  if (err) throw err;

  ifft(freq, restored, function (err2, out) {
    if (err2) throw err2;
    // out === restored
    // restored is approximately equal to time
  });
});

For both functions:

• input: Float32Array containing interleaved complex values [re0, im0, re1, im1, ...]
• output: Float32Array to receive output values in the same layout
• callback: (err, result) called asynchronously when done

Rules:

• input and output must be Float32Array
• input.length === output.length
• length must be even (pairs of real/imaginary values)
• number of complex samples (length / 2) must be a power of two

Test

npm test