Reply to thread

Message: [QUOTE="Bombastus, post: 91018, member: 2178"] FRCTL Audio specifies DLAY for up to 64 taps per channel, which means its stereo engine can expose 128 independent tap slots when unlinked. That figure is easy to miss because the product is usually described simply as a 64-tap delay, including in much of its launch coverage. The difference starts with stereo linking. In linked mode, DLAY uses 64 taps mirrored across the left and right channels, so a tap placed at one time position belongs to the shared stereo pattern rather than becoming a separate left-side and right-side event. Unlink the channels and the arithmetic changes. You get 64 taps on the left and another 64 on the right, with both sides operating independently for a maximum of 128 total taps. [HEADING=2]Stereo linking changes the meaning of 64 taps[/HEADING] A headline number like 64 taps can describe two very different working arrangements. With DLAY linked, you are editing one 64-tap structure that is mirrored across both channels, which is useful when the timing architecture should remain coherent from left to right. That does not mean every repeat must sound centered. DLAY still exposes per-tap pan and other tap-level shaping, but the underlying linked arrangement remains one shared set of 64 tap positions rather than two unrelated banks. The [B][URL='https://goldmidi.com/community/threads/frctl-audio-delivered-dlay-1-2-0-update.76246/']DLAY 1.2.0 multi-tap controls[/URL][/B] sit on top of that architecture. The important distinction for larger patterns is that stereo linking determines whether you are working with one mirrored bank or two independent channel banks. Unlinked mode is where DLAY becomes a 128-tap system in practical terms. Left can hold one set of 64 events while right holds another, so the timing, density, and treatment of one channel do not have to copy the other. Linked mode still has a strong practical role. If the musical idea depends on one repeat pattern staying synchronized across the stereo field, a shared bank is simpler to edit and leaves fewer independent events to manage when the arrangement changes. [HEADING=2]Unlinked banks create more than extra density[/HEADING] The useful part of 128 total taps is not simply that you can pile on twice as many echoes. Two [B]independent left and right tap banks[/B] let you allocate complexity where the arrangement needs it instead of forcing both channels to share the same event map. A vocal effect, for example, could use a restrained left pattern and a busier right pattern while preserving deliberate timing differences between them. A sound-design patch could separate short, clustered events on one side from longer-spaced events on the other without treating stereo as a single mirrored grid. That is different from taking one delay pattern and widening it afterward. The distinction happens at the tap-layout level, because each unlinked channel has its own capacity rather than relying on one common set of tap positions. You also do not need to chase 128 active taps for the mode to matter. Even modest patterns benefit when the left and right sides need different counts, different spacing, or different edits, since unlinking removes the requirement that both sides inherit the same basic structure. The editing consequence is easy to overlook. An unlinked patch can carry two separate rhythmic structures at once, so changing one side does not force the other side to follow the same event placement, which matters when you are building deliberately uneven stereo motion rather than simple alternating repeats. [HEADING=2]Maximum tap count does not equal maximum CPU cost[/HEADING] Tap count alone is also a poor shortcut for predicting processor load. FRCTL says clean taps are extremely light, while active taps using pitch or formant processing take the heavier path through Signalsmith Stretch. That means two sessions with the same number of active taps can place different demands on the computer. A dense pattern built mostly from clean repeats should not be assumed to cost the same as an equally dense pattern that performs pitch and formant processing across many active taps. That changes how you should test a heavy patch. Before cutting the tap count, disable pitch or formant on the busiest taps and compare the load, because FRCTL identifies those processes rather than ordinary clean repeats as the expensive path. FRCTL's own guidance is to use as many taps as the project needs and bounce when a crowded session becomes costly. That advice makes more sense once tap processing is separated from tap capacity, because freezing or printing a complex effect can recover headroom without forcing you to redesign the stereo pattern. The practical limit therefore depends on what those taps are doing, not only how many slots are filled. If a large unlinked design starts becoming expensive, the first thing to inspect is the processing applied across the active taps rather than treating 128 as an automatic ceiling your machine either handles or fails. [/QUOTE]

Name