From Sample & Hold to Sample-Rate Converters (2)
When you accelerate a step sequencer's clock to audio rates, the stepped output becomes a new waveform. Push further and you cross into digital audio territory, where sample rate determines quality and aliasing artifacts appear when the rate is too low.
Read the original article on Sound On Sound →Two Halves of Digital Audio
Converting analog sound to digital requires two steps, each with its own kind of degradation:
Sampling captures the signal at regular intervals: the sample rate. The Nyquist theorem says you need at least twice the highest frequency to reconstruct it faithfully. Drop below that and you get aliasing, phantom frequencies that fold back into the audible range. CD audio uses 44,100 samples per second, capturing everything up to ~22 kHz.
Quantization rounds each sample to the nearest available level, the bit depth. 16-bit audio has 65,536 levels and a noise floor around −96 dB. Drop to 8 bits (256 levels) and you hear gritty noise. At 4 bits (16 levels) the signal is barely recognizable. At 2 bits you get a harsh square-like approximation.
Try it: start with CD Quality, then drag the sample rate down to hear aliasing appear. Switch to the quantization side and crush the bit depth to hear the noise floor rise.
References
- Nyquist–Shannon sampling theorem, Wikipedia
- Quantization, Wikipedia