1. Build a signal from waves
Adjust amplitudes and frequencies. The signal below is a linear combination of sine waves.
2. Frequency spectrum
The spectrum shows how much each frequency is present. Peaks should appear near the frequencies you selected above.
3. Low-pass filtering
A low-pass filter keeps slow waves and removes fast waves. Try changing the cutoff.
4. Compression by keeping coefficients
Fourier compression keeps the largest wave coordinates and discards the rest.
5. Sharp edges need many frequencies
Smooth signals compress well. Signals with jumps need many frequencies. This is why edges and discontinuities are expensive in Fourier language.
6. Tiny image and 2D frequency intuition
This synthetic image contains horizontal and vertical wave patterns plus a bright object. The right panel shows a simplified Fourier magnitude picture.
7. Chapter summary
Time language
Coordinates are sample values: what happens at each time.
Frequency language
Coordinates are wave strengths: which frequencies are hidden inside.
Linear algebra
The DFT is a matrix transformation. The FFT computes it efficiently.
Applications
Filtering, compression, denoising, audio, images, PDEs, and AI features.