Sample rate is the resolution of time in digital audio: 44,100 measurements per second on CDs, 48,000 in most production and video work, higher in some studios. Each rate captures frequencies up to half itself, its Nyquist frequency.
How sampling works
Digital audio freezes the waveform’s voltage tens of thousands of times a second and stores the numbers. The sampling theorem guarantees the trade exactly: any frequency below half the sample rate is captured perfectly, anything above it folds back as aliasing unless filtered out first. So 44.1 kHz covers audio up to 22.05 kHz, comfortably past the roughly 20 kHz ceiling of human hearing. Bit depth is the other axis, how finely each measurement is graded, and the two together define digital fidelity.
The common rates
- 44.1 kHz: the CD standard, still the default for music distribution
- 48 kHz: video, film, streaming production, most live work
- 88.2 and 96 kHz: studio capture, where the headroom helps processing rather than hearing
- 22 kHz and below: lo-fi by choice, samplers, and the crunch of early digital
The odd 44.1 number is history frozen into a standard, chosen when digital audio was stored on adapted video equipment, and it persists because it works.
Sample rate in real-time systems
For real-time systems the rate also ticks the clock: latency budgets, buffer sizes, and analysis windows are all measured in samples, so the rate converts every engineering decision into milliseconds. An FFT of 1024 samples spans 23 ms at 44.1 kHz and 21 ms at 48 kHz; the same code gets slightly faster and slightly less frequency-precise as the rate climbs. Analysis systems like Photism’s care about the rate mostly through this arithmetic.