A voice note begins as a physical event: your vocal cords create pressure waves in the air. To store this on a device, that analog wave must be converted into a series of numbers. This process is called digitalization. The most common method for this is Pulse Code Modulation (PCM), which is the raw, uncompressed format you find in WAV files. PCM works by taking snapshots, or samples, of the analog audio wave at thousands of times per second.

Two key parameters define the quality of this conversion: the sampling rate and the bit depth. Let's start with the rate.

The sampling rate is how many snapshots are taken per second, measured in Hertz (Hz). For a CD, the rate is 44,100 Hz. For professional recording, it can be 48,000 Hz, 96,000 Hz, or even higher. The correct rate is dictated by the Nyquist-Shannon sampling theorem.

This theorem states that to accurately capture a frequency, you must sample at a rate at least twice that frequency. The range of human hearing tops out around 20,000 Hz, so a rate of 44,100 Hz provides a comfortable margin to capture everything we can hear. Since most of the crucial information in human speech lies well below 10,000 Hz, standard voice note apps can use even lower rates.

Failing to follow this rule creates a type of distortion called aliasing, where high frequencies that weren't sampled fast enough get misinterpreted as lower frequencies. It's like seeing a car's wheels appear to spin backward in a video; the camera's frame rate isn't high enough to capture the true motion. To prevent this, Analog-to-Digital Converters (ADCs) use an anti-aliasing filter to chop off any frequencies above the recordable limit before sampling begins.

While sampling rate determines how often you measure the wave, bit depth determines how precisely you measure its amplitude (or volume). Each sample is assigned a value from a fixed set of possible levels. Bit depth defines how many levels are in that set.

A higher bit depth provides more possible levels, resulting in a more accurate measurement. A 16-bit recording has 65,536 levels, while a 24-bit recording has over 16.7 million. This precision directly impacts the dynamic range—the difference between the quietest sound you can capture and the loudest sound before distortion. More bits give you a wider dynamic range and a lower noise floor—the hiss of inherent system noise.

The small difference between the actual analog value and the quantized digital value is called quantization error. With low bit depths, this error can become audible as a grainy noise. A higher bit depth minimizes this error, leading to a cleaner recording.

Now that we have a clean, uncompressed digital signal, the next step is to make it a practical size for sending and storing.