Converting continuous analog audio into discrete digital data points — determines later sound quality and file size. Foundation of every digital recording.
On set or in post-production, we encounter sampling daily — often without much thought. The sound a microphone captures is continuous, analog. To bring it into the digital realm, this flowing stream must be "sampled" at regular intervals — meaning, snapshots are taken. These measurement points are called samples. The more frequently we sample per second, the more accurately we represent the original waveform.
The sample rate — measured in Hertz or Kilohertz — determines the foundation. 48 kHz is the standard in film: we therefore take 48,000 measurements per second. The Nyquist theorem dictates this: the highest frequency we can accurately capture is half the sample rate — thus, a maximum of 24 kHz. This is more than enough for human ears (we hear up to about 20 kHz), but: sampling below 48 kHz — for example, 44.1 kHz as with CD audio — noticeably loses brilliance at the high end. Conversely, 96 kHz offers little advantage in dramatic film; however, the data volume increases significantly.
In practice, errors become immediately apparent if sampling is done incorrectly. This phenomenon is called aliasing: high-frequency tones that exceed the Nyquist limit are misinterpreted and create disturbing artifacts — metallic hums, unexplained buzzing. Therefore, audio interfaces employ an anti-aliasing filter that limits incoming frequencies. This has long been standard in editing; there, the sample rate determines the precision with which we can calculate effects, pitch shift, or time-stretch.
The second dimension is the bit depth or word length: 16 bit (CD standard) or 24 bit (professional). This determines how finely we can quantize the amplitude of each sample — how many volume steps exist between silence and maximum. 24 bit provides us with 16 million steps, 16 bit only 65,536. In dialogue recording, especially with weak or dynamic sources, 24 bit pays off: more headroom, finer control in post-processing, less audible quantization noise.
For sync or multi-track recording, the rule is: all tracks must run at the same sample rate. A mix of 48 kHz and 44.1 kHz material leads to phase errors and timing problems that are difficult to repair. Desirers, plugins, and DAWs — all operate at this rate. Only in the final mixdown or during rendering to the output format can conversion occur, but even then, some transparency is lost.