See sound
Pick a sound, or turn on your microphone. Time runs left to right, pitch runs bottom to top, and the brighter a spot is the louder that pitch was at that moment. Point at the picture to read exact numbers.
Pick a sound
One note at 196 Hz, with overtones stacked evenly above it. That stack is why it sounds like a guitar and not a flute.
Detail
Balanced · 2048 samplesThis is the tradeoff, and it is the whole lesson. Sharper timing blurs the pitch; finer pitch blurs the timing. Nothing makes both sharp at once.
View
Linear is what the FFT returns, log matches how pitch is heard, mel matches how hearing spaces it out.
Everything else
How many samples go into one column. This is what the detail control sets.
How much each window shares with the last one. It sets the hop.
Which slice of dB the colour ramp is stretched across.
Hann — cosine taper, the usual choice
Cutting a signal off abruptly invents frequencies that were never in it — leakage. A taper is how that is kept down.