Reply to thread

Message: [QUOTE="Bombastus, post: 91999, member: 2178"] At 44.1 kHz, alias components below a sawtooth fundamental can remain audible at levels that higher aliases may hide behind nearby harmonics. Aliasing therefore has no single signature sound. Sometimes it is an obvious extra tone; sometimes it is a thin layer of grit you barely notice until it disappears. A useful ear test starts with movement, not adjectives such as harsh, fizzy, or metallic. Once you understand the [B][URL='https://goldmidi.com/community/resources/what-is-audio-aliasing.61/']frequency folding behind audio aliasing[/URL][/B], listen for tones that move in the wrong direction as pitch rises, because clean harmonic content climbs with the note while aliases can turn around and descend. The effect becomes easiest to spot on sparse material. A bright synth note, a driven sine sweep, a clipped high lead, or a heavily distorted guitar signal gives the unwanted components room to poke through. Dense drums, cymbals, reverb, and layered instruments can bury the same artifacts surprisingly well. [HEADING=2]Moving tones expose aliasing better than harshness[/HEADING] Run a sine sweep through a nonlinear plug-in with oversampling disabled and listen above the midrange. Harmonics created by the processor rise with the sweep until they cross the Nyquist limit, then fold back into the audible band. Your source keeps climbing while some new tones begin falling. Reverse movement is far more diagnostic than a vague digital edge. A saturator can sound bright without aliasing, and a clipper can sound abrasive because clipping is supposed to create harmonics. An alias becomes easier to identify when its pitch travels independently of the source or reverses direction while the source continues upward. Repeated folding can make several tones seem to weave through one another. A strong distortion stage may create third, fifth, seventh, and higher harmonics, each reaching Nyquist at a different moment, so you hear shifting non-harmonic whistles, chirps, or rough tones rather than one permanent layer of fuzz. High synth notes expose the problem for a similar reason. Saw and pulse waveforms already contain many harmonics before they reach a filter or distortion stage. As the fundamental rises, less room remains between those harmonics and Nyquist, so poorly band-limited oscillators can produce audible components that no longer track the musical harmonic series. [HEADING=2]Audibility changes with pitch and masking[/HEADING] Aliasing can also sound quieter than it looks. A spectrum analyzer may show plenty of folded components while your ears barely register them because nearby legitimate harmonics mask the intruders. Judging an analyzer trace by height alone can make a small technical flaw look enormous. Perception is not uniform across the spectrum either. The [B][URL='https://doi.org/10.1121/1.4748964']measured audibility of aliased sawtooth components[/URL][/B] changes with fundamental pitch and with whether the unwanted energy sits above or below it. Components below the fundamental can receive less useful masking from the wanted tone, which makes some low aliases stand out even when their level looks modest. Aliasing can therefore sound like an unrelated low whistle rather than extra brightness. A high harmonic may fold far below the note that generated it, where the ear no longer groups it neatly with the original harmonic structure, especially on sustained notes with a stable pitch. Full mixes change the picture again. Many weak aliases spread across different frequencies instead of stacking into one obvious whistle. The result can become a faint roughness, grain, or low-level haze, and turning oversampling on may clean the texture without producing a dramatic before-and-after moment. [HEADING=2]A controlled listening test separates the artifact[/HEADING] Use a clean sine tone or slow sine sweep when you need certainty. Keep the input level fixed, turn off modulation and time-based effects, then compare the suspect processor with oversampling disabled and enabled at matched output level. Any disappearing side tones deserve attention. Level matching matters because oversampling modes can alter gain slightly in some processors. A louder version often wins a casual comparison even when the artifact level is worse. Keep your monitor level steady too, since faint aliases become much easier to notice when you simply turn the whole test up. Try the same processor at 44.1 or 48 kHz, then repeat at 88.2 or 96 kHz without changing the musical pitch. Aliasing products should move because the Nyquist boundary moved, while ordinary harmonic distortion below Nyquist stays tied to the source. A final test uses real material rather than a laboratory sweep. Play exposed high notes, bright transients, or sustained distorted tones and switch oversampling while listening for descending whistles, unstable high-frequency grain, or low tones that do not belong to the chord. If the analyzer changes but the track does not, the visible aliasing may simply be too well masked to matter. [/QUOTE]

Name