Want to know how aliasing sounds?

[Guillermo Luijk]

Aliasing is a common issue in any digital sampling system (a digital camera), taking place when trying to sample (taking a picture of) some signal which has frequency components (fine detail) above the sampler (sensor) limits. Luckily for photographers, aliasing in the image world is seldom a big problem, that's why camera manufacturers at some point even decided to start renouncing to implement AA filters on their cameras. By doing so we collect some extra fine detail from the lens at the price of (maybe) suffering some aliasing artifacts under certain circumstances.

The technical reason for aliasing is that if you sample your signal at an inssuficient rate (not enough Mpx), the frequencies beyond the capabilities of the sampler come back to you in the form of a high frequency interference which can be interpreted as noise. In the imaging world this interference can be enhanced by fooling the Bayer demosaicing algorithm producing colour artifacts (moiré), but it is not only Bayer sensors that can suffer aliasing, Foveon sensors without proper AA filtering will also be prone to aliasing artifacts, but in this case much less harmful since they will be monochrome.

Unluckily for the audio world, aliasing in a sound is very difficult to hide to the ear, and moreover it can easily take place not only in the A/D (i.e. sampling) stage, but very easily during sound processing. The only way to suffer from aliasing on image processing is during rescaling/resampling.

This is a very short video showing how a strongly aliased sound looks like and plays, comparing its waveform and spectrum with and without aliasing. The non-aliased sound is not pleasant itself for being such a high frequency sawtooth waveform, but you'll agree the aliased version being much worse (keep your volume moderate when playing):

https://www.youtube.com/watch?v=vIEHxinB134

Regards

[JaapD]

Most of us, me included  , have a kind of obsession for the ‘sharpest available lens’. Be aware though that the sharpest lens at the same time provides the largest portion of aliasing in the image.

As an example, if your lens has twice the bandwidth that your sensor can handle you’ll get aliasing all over the place, folding back from the high frequency details up till the low frequency structures.

Luckily for us in real life the right side of the lens’s MTF curve is already significantly down, acting as a kind of optical low pass filter.

Regards,
Jaap.

[Guillermo Luijk]

In addition to that, unless we are talking about periodic patterns, aliasing on images (for example natural textures) will render in some form of noise, which can increase the perception of sharpness not being harmful as in the audio world. It is not real detail, but it doesn't make image quality worse.

Regards