Reply to thread

Message: [QUOTE="Bombastus, post: 91693, member: 2178"] Quadrant Modulator’s pitch-follow mode uses a phase-locked loop to drive its internal oscillator from the main input or side-chain signal. On a clean monophonic note, the relationship can settle into something musically useful. Feed the tracker a chord, noisy transient, or weak fundamental and the carrier can move for reasons you did not play. Pitch-following ring modulation sounds simple because the carrier follows the note instead of sitting at one fixed frequency. The important detail is what happens after a tracking error. Every carrier movement immediately relocates the ring modulator’s sum and difference sidebands. A [B][URL='https://goldmidi.com/community/threads/unusable-engineering-introduced-quadrant-modulator.77074/']side-chain pitch-following ring modulation path[/URL][/B] therefore behaves differently from an ordinary pitch meter that briefly shows the wrong note. A bad estimate becomes audible modulation, often before you have time to notice the tracker itself was uncertain. [HEADING=2]Carrier errors move both sidebands at once[/HEADING] Suppose one source partial sits at 500 Hz and the tracked carrier should be 200 Hz. Ideal ring modulation places the new components at 300 Hz and 700 Hz. If the carrier suddenly jumps to 220 Hz, those components move to 280 Hz and 720 Hz instead. One tracking error has pushed the two sidebands in opposite directions. Repeated small errors create moving spectral edges, while a bigger octave mistake can reorganize the effect in a single jump. You hear a chirp, scrape, or metallic pitch lurch rather than a neat correction from one detected note to another. Pitch tracking also changes the musical value of carrier ratios. When the carrier is tied consistently to a simple fraction of a monophonic fundamental, the generated components can stay on a repeatable frequency grid. Let the estimate wander and the grid moves with it, so previously stable relationships begin beating against the source and against earlier resonances in the signal chain. Published [B][URL='https://www.isca-archive.org/eurospeech_2003/pelle03_eurospeech.pdf']research on PLL pitch estimation under difficult signals[/URL][/B] identifies pitch doubling, pitch halving, noise degradation, and errors around voiced segment boundaries as recurring detection problems. Ring modulation makes those mistakes unusually exposed because frequency estimation is not merely displayed or converted to MIDI. It becomes the oscillator controlling the multiplication itself. [HEADING=2]Transients and chords give the tracker conflicting evidence[/HEADING] A plucked string is not born as a steady periodic waveform. Its first milliseconds can contain pick noise, broadband attack energy, changing partial levels, and a fundamental that becomes clearer only after the transient settles. A pitch follower may therefore spend the attack moving toward a stable estimate rather than arriving there instantly. Low notes add another practical limit because one waveform period takes longer. Forty hertz lasts 25 milliseconds per cycle, while 200 Hz completes a cycle in 5 milliseconds. Any detector that needs repeated periodic evidence has less new pitch information available during the opening instant of the lower note. Polyphonic material is harsher. A monophonic pitch tracker has one output value, while a chord presents several fundamentals and overlapping harmonics at the same time. The tracker can only produce one carrier estimate, so changes in note balance may pull the result toward a different periodic component even though nobody changed the chord voicing. Octave errors are especially destructive here. If a weak fundamental lets a stronger second harmonic dominate the estimate, the detected carrier can effectively jump to twice the intended relationship. Ring-modulated sidebands then move by the carrier-frequency difference, which can turn a controlled tuned texture into a much brighter and more inharmonic burst. [HEADING=2]Cleaner control signals make pitch following more predictable[/HEADING] The best source for the effect is often not the richest source in the mix. A dry vocal, DI bass, isolated synth oscillator, or filtered side-chain can give the tracker clearer periodic information than a distorted bus carrying the sound you actually want to process. You can keep the complex signal in the audio path while giving pitch detection an easier job elsewhere. Input conditioning matters for the same reason. Heavy distortion before detection creates extra harmonics, noise can blur periodic structure, and long ambience can leave old pitched information hanging underneath the new note. Moving those processes after the tracked ring modulator can make the carrier behave more deliberately without making the final sound polite. Fast playing exposes another compromise. A tracker that reacts aggressively can follow note changes quickly but may also chase short irregularities, while stronger smoothing can resist jitter and take longer to settle after a genuine pitch move. PLL implementations expose versions of this trade-off through loop filtering, feedback, frequency range, and lock behavior. On guitar or bass, split the signal before fuzz. Send the clean branch to the side-chain and the distorted branch through the effect, so pitch detection sees a clearer fundamental while the multiplier still receives the harmonically dense tone. The carrier then follows the note source rather than the extra partials created by the distortion. [/QUOTE]

Name