the manual
Learn
Signal processing, one idea at a time. Each chapter grew out of a panel in the playground and keeps the same live interactive to poke at. Read it top to bottom, or skip straight to whatever you're curious about.
- 01Sound & samplingA microphone turns a pressure wave into numbers. What amplitude is, what a sample actually captures, and why the rate matters.
- 02Nyquist & aliasingWhy the sample rate sets a ceiling on the frequencies you can capture, and what happens when you cross it.
- 03FFT & windowingSamples become a spectrum through the FFT, at a price: you can know a frequency precisely, or know it soon, not both.
- 04Harmonics & timbreA musical note is never just one frequency. The overtone stack riding on top is what turns a plain tone into someone in particular.
- 05Peak detectionFinding the frequencies that matter in a noisy spectrum, and refining them below one bin with parabolic interpolation.
- 06Feature extractionEvery analysis window collapses into the handful of streams a recognizer actually looks at: spectrum, peaks, envelope.
- 07Building a recognizerTurning a feature stream into glyphs, from segmentation through confidence to the plugin contract that ties it together.
- 08DTMF: history & why it worksThe dual-tone scheme wired into every phone keypad, and the deliberate choices that keep it readable through noise and crosstalk.
- 09FFT vs. GoertzelThe FFT measures every frequency at once; Goertzel only asks about the ones you care about. General-purpose versus purpose-built.