kvantiseringstøy

Kvantiseringstøy, also known as quantization error or quantization noise, is an inherent artifact that arises when a continuous analog signal is converted into a discrete digital signal. This process, called analog-to-digital conversion (ADC), involves sampling the analog signal at regular intervals and then assigning each sample a discrete numerical value from a finite set of possible values. The precision of this digital representation is limited by the number of bits used to encode each sample.

The difference between the actual analog value of a sample and the closest available discrete digital value is the quantization error. This error can be positive or negative and is often treated as a random noise source. The magnitude of the quantization error is typically bounded and depends on the resolution of the ADC. A higher resolution (more bits) results in smaller quantization steps and thus lower quantization error.

Quantization noise can manifest as undesirable artifacts in digital audio, images, and other digitized data. In audio, it might be perceived as a grainy or distorted sound, especially at low signal levels. In digital images, it can lead to banding or contouring in areas of smooth gradients. Various techniques, such as dithering, can be employed to mitigate the perceptual effects of quantization noise by intentionally introducing a small amount of random noise that spreads the quantization error across a wider frequency range, making it less noticeable.