FusionDesk User Guide

1. Introduction

1.1 What Is FusionDesk?

FusionDesk is a software audio mixer for Windows, built for people who take sound seriously.

Whether you're a home studio enthusiast routing headphone mixes, a musician managing stage monitors, a DJ crafting a pre-listen mix before it hits the PA, or an engineer running livestreams or corporate events — FusionDesk is designed to give you the kind of routing power and flexibility that until now has only existed in dedicated hardware or expensive, specialised software.

The inspiration is RME's TotalMix FX: a professional-grade mixing matrix that ships with RME audio interfaces. TotalMix is brilliant, but it's locked to RME hardware. FusionDesk takes that philosophy and opens it up — to any audio device on your system, from professional ASIO interfaces down to standard Windows audio devices and Windows applications.

1.2 Who Is It For?

Home studio enthusiasts who want more than Windows' basic audio routing without the complexity of virtual cable stacks.

Musicians managing personal monitor mixes — hear what you need to perform at your best, independently of the front-of-house mix.

DJs who want a dedicated pre-listen and cue mix before a track goes to the room.

Live sound engineers who want a digital console's routing, processing and monitoring in software — and, on a touchscreen, a hands-on mixing surface.

Livestreamers and content creators managing multiple audio sources — music, microphone, game audio, guest feeds — with precise control over what goes where.

AV professionals integrating software routing into installs alongside matrix hardware.

1.3 The Problem It Solves

Windows audio routing is surprisingly limited. You can select a default input and output device. That's about it.

If you want to:

• Send different mixes to headphones vs. speakers
• Route a music player to one output while your DAW goes to another
• Run a monitor mix independently from your main output
• Blend ASIO hardware channels with system audio sources
• Capture audio from individual applications

…you're reaching for third-party tools like Voicemeeter, which are powerful but complex, or paying for hardware matrix routers.

FusionDesk sits in between: the routing flexibility of a hardware matrix, the accessibility of software, and a UI that won't intimidate an enthusiast but won't embarrass a professional.

2. Getting Started

2.1 System Requirements

• Operating system: Windows 10 (version 2004 or later) or Windows 11, 64-bit
• Per-Application Audio Capture: Windows 11 or Windows Server 2022 only
• Audio hardware: Any ASIO driver or standard Windows audio device
• RAM: 16 GB recommended; more if using long loop recording across many channels (see Loop Recording)

An ASIO driver is recommended for low-latency, professional-quality operation. Windows Audio (WASAPI) shared mode is supported for all other devices.

This short walkthrough takes you from a fresh install to hearing your first mix and saving it — the fast path to getting sound flowing.

2.2 Install and Launch

Run the FusionDesk installer and follow the on-screen steps, then launch FusionDesk. The installer places the application in Program Files and includes a copy of this guide and the Known Issues document.

FusionDesk enforces a single instance — if it is already running, a second launch brings the existing window to the front. A system tray icon is available via Menu → Window → Show Tray Icon; right-clicking it gives quick access to show, hide, or exit.

On first launch you'll see 8 input strips across the top and 8 output strips along the bottom, with the Control Room panel on the right. No devices are assigned yet — each channel shows a blank selector at the top. The first output strip is labelled Main, the second Monitor.

2.3 Choose Your Audio Device

FusionDesk can use professional ASIO interfaces or standard Windows audio devices — or both at once.

• If you have an ASIO interface, click the ASIO driver selector at the top of the Control Room panel and choose it. This sets the device used for all ASIO input and output strips. Set its sample rate from the Control Room.
• If you don't, skip the ASIO selector entirely — Windows Audio (WASAPI) devices such as USB headsets, HDMI outputs, and Bluetooth speakers are available directly from each channel's own selector. (Alternatively, the free ASIO4ALL v2 can wrap any Windows device as an ASIO driver.)

Individual channels can use any combination of ASIO channels, Windows Audio (WASAPI) devices, or per-application capture sources — see Chapter 4 for full details.

2.4 Assign an Input and an Output

1. Click the device selector at the top of an input strip and pick a source — an ASIO channel, a microphone, or a playback device.
2. On the Main output strip, click its device selector and choose your speakers or headphones.

Tip — per-application sources: just play audio in any app (a browser, a media player) and it appears automatically at the bottom of every input strip's selector. Pick it like any other source. (Requires Windows 11.)

2.5 Make Sound

1. Raise the input fader. All faders start fully down (silent) by design, so nothing plays at full volume by accident.
2. Raise the Main output fader.

You should now hear your input through the Main output.

The strip faders you just raised feed whichever Mix is selected — Mix 1 by default, which goes to Main. Each Mix button at the bottom of a channel is a separate output destination; select a different one to send the same (or different) inputs to a different output. That's the whole idea — source(s) → input strip(s) → mix → output.

Stuck? Hover any control and a one-line description appears in the title bar at the top of the window. If a channel shows no signal, check its device selector is assigned and the fader is up.

2.6 Save It as a Snapshot

Once you have a setup you like, save it so you never have to rebuild it: open the main menu — the ≡ button in the top-right of the Control Room — then choose Snapshot → Save.

A snapshot captures your complete mixer state: every device assignment, fader, pan, route, mix, and plugin. Recall it any time to return to exactly this setup.

From here: build more mixes (§6), add processing (§9), play instruments with MIDI (§4), and record (§11).

3. The Interface

3.1 Layout Overview

FusionDesk's layout follows the TotalMix FX convention:

• Input strips — a row across the top. One channel per audio source.
• Output strips — a row across the bottom. One channel per audio destination.
• Control Room panel — fixed on the right. Global controls, monitoring, system status.

The number of visible channels is configurable from 4 to 32. Window width scales automatically to fit. Use the hamburger menu (≡, top-right of the Control Room) → Channel View to change the count.

FusionDesk — 12 input strips (Windows applications, hardware microphones, and MIDI instruments side by side), 12 output strips, and the Control Room panel. A VST3 plugin, An EQ, a compressor, and a limiter are loaded on the Monitor output. The active mix (or mixes) are highlighted in the Mix bar at the bottom.

Channel View and Bank View

There are two ways to view your channels, chosen from the hamburger menu (≡):

• Channel View — all channels in a single row, 4 to 32. This is the default. Pick the channel count here.
• Bank View — for sessions with more channels than fit comfortably across your screen, the channels are split into equal banks (pages) that you step through one at a time. Choose a preset such as 2 Banks × 16 Channels or 4 Banks × 8 Channels — the first number is how many banks, the second is how many channels each bank shows.

When Bank View is active, a bank selector appears on the Control Room's Mix row:

• ◀ / ▶ — go to the previous / next bank (also the Left and Right arrow keys).
• Bank n of m — click to jump straight to any bank (also Alt + 1-4).

Inputs and outputs always show the same bank together, so the Mix numbers line up across both rows. Switching banks changes only what is on screen — every channel keeps running and routing in every bank, so a mix in a bank you are not currently looking at is still live. Your chosen view is remembered between sessions and saved with snapshots.

To rearrange channels while in Bank View, switch to Channel View, drag them into order, then switch back — the order carries over. (See Known Issues — moving a channel directly between banks is not yet available.)

3.2 Input Strips

Each input strip contains, from top to bottom:

• Device selector — choose the audio source for this channel
• VST slots — up to 6 insert plugin slots (click the channel name to show them in the Control Room)
• Gain, Pan, Width sliders
• Stereo level meters with the fader between them
• dB readout — shows current fader level; double-click to type an exact value
• RMS peak indicators (bottom of channel)
• dBTP peak indicator (top of channel)
• Signal presence indicator — a small light at the top of each channel. Green when audio is present, yellow for MIDI activity, red when the signal has clipped. See Section 7.3
• Mute / PFL / AFL / Solo / Record buttons
• Mix buttons — select which mix this fader belongs to

3.3 Output Strips

Output strips mirror input strips, but represent audio destinations rather than sources. The fader is the master level for that Mix. Output strips also have up to 6 VST insert slots (click the channel name to show them in the Control Room).

The first output strip is always Main (your primary output). The second is always Monitor (your control room / headphone output). Their names are fixed and non-editable.

3.4 The Control Room Panel

The Control Room panel is fixed at the right edge of the window. It contains all global controls, metering, monitoring, and recording functions.

3.4.1 Audio Load & Device Info

The top of the panel shows the Audio Load and device information — bit depth, buffer size, and in/out latency. These update live as the engine runs. See Appendix F for how these values are calculated.

The Sample Rate row shows the current sample rate as a dropdown. When an ASIO device is active, click it to change the sample rate — the driver restarts immediately. The dropdown is greyed out when using Windows Audio, where the sample rate is set by Windows.

The dropdown lists only the rates the ASIO driver reports as available. Some devices — particularly live sound consoles — lock the sample rate in their own control software or hardware, in which case only one rate will appear and selecting it has no effect (the device is already running at that rate and will not allow an external change). If another application has opened the same ASIO driver, it controls the sample rate; FusionDesk will reflect whichever rate the driver is actually running at.

3.4.2 LUFS Display

Shows short-term (3 s) and momentary (400 ms) loudness for the monitor output, following EBU R128. Two Max buttons alongside latch the highest values reached — Peaks Max for the meter peaks, LUFS Max for the loudness values. See Section 7.4.

3.4.3 ASIO Driver Selector

Choose the active ASIO driver. All ASIO channel pairs from the selected driver become available in every channel's device selector. Select None if you are using Windows Audio devices only. See Section 4.1.

Reconnect Current — at the top of the ASIO dropdown (below None). Use this when your interface was powered on after FusionDesk launched and the app fell back to Windows Audio. Picking Reconnect re-opens the currently-selected ASIO driver without needing to switch to None and back. The item is greyed out when no ASIO driver is selected.

After a PC reboot, FusionDesk may not auto-reconnect to the ASIO driver even if the same driver was active before. Use Reconnect Current to restore the connection. (After simply closing and reopening FusionDesk without a reboot, auto-reconnect normally works.)

3.4.4 Monitor Listen Output Selector

Assigns a separate physical output for PFL/AFL/Solo monitoring, independent of the Monitor channel's output. When a listen output is assigned, the Monitor mix continues uninterrupted on its own output while listen signals play on the listen output. The active selector is shown in green. See Section 8.6.

3.4.5 Arm & Exclusive Buttons

Five Arm buttons and five Exclusive buttons control how channel buttons behave across all channels.

Arm buttons enable the corresponding function on channels. With an Arm button off, the corresponding channel buttons do nothing — useful for preventing accidental changes during a performance:

Exclusive buttons enforce a one-at-a-time rule for the corresponding function. When active, engaging one channel's button clears the same button on all other channels:

A channel's PFL and AFL/Solo are always mutually exclusive — turning one on switches the other off — since you can only listen one way at a time. This is independent of the Exclusive buttons above. The Monitor channel's own PFL button is exempt, acting as an independent A/B toggle for the monitor VST chain.

3.4.6 VST Slots

When a channel name is clicked, its VST insert chain appears here — up to 6 slots per channel. See Chapter 9.

3.4.7 Recording Controls

Start, stop, split, and capture recordings; select the recording format; set the loop length; and monitor recording progress. See Chapter 11.

3.4.8 Hamburger Menu (≡)

Access to channels, app sources, snapshots, channel-strip actions, window options, settings, and help. Click the ≡ icon at the top right of the Control Room panel. Click it again, press Escape, or click anywhere outside the menu to close it. Use the Up/Down keys to navigate items, Left to open a submenu, and Right to go back to the parent menu.

3.4.9 Control Hints

Control hints appear in the title bar when you hover a control, including keyboard shortcut reminders. Turn them on or off via Menu → Window → Show Tooltips.

3.5 Channel Names & Colours

Double-click a channel name to edit it. Right-click the name to open the colour menu — pick one of the ten preset colours, copy/paste a colour, or choose Custom Colour... for a full colour picker (HSV wheel, sliders, or hex input). Useful for visual grouping: inputs from one performer, outputs to one destination, drums vs vocals, etc.

Main and Monitor output names are fixed and cannot be edited. Use Channel Strips → Clear Channel Names to reset visible names and colours.

3.6 Reordering Channels

Drag any channel name horizontally to reorder channels. Input and output rows reorder independently. Visual position is completely independent of the underlying routing — rearranging channels never changes what you hear.

Main and Monitor are pinned. Main and Monitor are pinned to one edge of the output row — Main always sits on the outer edge with Monitor immediately inside it. They cannot be dragged out of position, and no other output strip can be dropped into their slots. This keeps the master output and the PFL/AFL listen path always visible regardless of channel count.

The pinning side is set under Settings → Behaviour → Main/Monitor Position:

• Left edge (default) — Main on the far left, Monitor just to its right.
• Right edge — Main on the far right (next to the Control Room panel), Monitor just to its left. In this mode the pinned pair slides automatically when you change the channel count, so they always occupy the rightmost two visible channels.

Switching the setting applies immediately — no restart required.

4. Device Sources

4.1 ASIO Hardware Channels

ASIO (Audio Stream Input/Output) drivers give FusionDesk direct, low-latency access to professional audio interfaces. Channels appear in the device selector as stereo pairs using the names reported by the driver — for example, "Analog (1+2)", "ADAT (3+4)".

Select your ASIO driver in the Control Room panel. All ASIO channel pairs from that driver are then available in every input and output strip's device selector.

Only one ASIO driver can be active at a time. Switching drivers is live — no restart required — but causes a brief audio interruption while the engine reinitialises.

4.2 Windows Audio (WASAPI) Devices

Any standard Windows audio device — USB headsets, HDMI outputs, Bluetooth speakers, virtual audio cables — appears in the device selector alongside ASIO channels. These devices use Windows Audio shared mode.

Windows Audio adds more latency than ASIO, but it works with any device and can be mixed freely with ASIO sources in the same session.

4.3 Per-Application Audio Capture (Windows 11)

FusionDesk can capture the audio output of individual applications — browsers, media players, games, communication apps — and route them into the mix like any other input. No virtual cables required.

Simply play audio in any application and it will appear automatically in every input strip's device selector. Select it, set your level, and FusionDesk remembers it for the next session. Full details in Chapter 14.

Requires Windows 11 or Windows Server 2022.

4.4 Virtual ASIO Drivers

Any ASIO-compatible virtual driver appears in the Control Room ASIO selector and can be used like any other ASIO interface. Two common examples:

• ASIO4ALL v2 (free, asio4all.org) wraps any Windows audio device — built-in PC audio, USB headsets, HDMI outputs — and presents it as an ASIO interface. Useful if you don't have a dedicated audio interface. For the lowest latency, a native ASIO driver on a dedicated interface is still recommended.
• Dante Virtual Soundcard and other network-audio drivers install an ASIO driver of their own and likewise appear as a standard ASIO device — so a Dante rig's channels are available in FusionDesk like any other ASIO interface.

4.5 Default Input / Default Output

Two special entries appear at the top of every device selector:

• Default Input — resolves to whichever device Windows currently treats as the default recording device
• Default Output — resolves to whichever device Windows currently treats as the default playback device

If your default device is an ASIO channel, FusionDesk will match it to the correct ASIO pair automatically. If it's a standard Windows Audio device, it opens via Windows Audio. Either way, the channel follows whatever Windows considers the default — useful for channels that should track the system default rather than being tied to a specific device.

If you change the default device in Windows Settings while FusionDesk is running, any channel using Default Input or Default Output switches to the new device automatically.

4.6 MIDI Input (Playing Instruments)

If you load an instrument plugin into an input strip, you can play it with MIDI. The input selector has a MIDI Inputs section, just under Default Input, listing:

• PC Keyboard MIDI — turns your computer's keyboard into a playable musical keyboard. The home row (A S D F G H J K L) plays the white notes, the row above plays the black notes, and Z / X shift down / up an octave.
• Any connected MIDI controllers — keyboards, pad controllers, and other MIDI devices plugged into your computer. Newly connected devices appear automatically the next time you open the selector.

Pick a MIDI source on the channel and play — the notes go to that channel's instrument plugin. The same source can be selected on several channels at once, and each plays independently, so one keyboard can drive several instruments. Selecting a MIDI source sets that channel's audio input to none (the channel's sound now comes from the instrument, not an audio device).

Set the host tempo for tempo-aware instruments in Settings → Behaviour → Tempo (BPM), range 20–300 BPM.

Organising instruments in the plugin browser. The plugin browser separates Effects Plugins from MIDI Instruments. FusionDesk makes a best guess when it first sees a plugin, but you can drag any plugin between the columns to suit how you actually use it — your arrangement is remembered.

Clearing a stuck note. If a note ever hangs, open the channel's input selector and click the same MIDI device again. This sends an all-notes-off to that channel and silences it.

5. Signal Controls

5.1 Gain

The Gain slider trims the input level before it reaches the mix. Range: −18 dB to +18 dB. Use it to calibrate source levels so your faders sit in a useful working range.

Scroll the mouse wheel over the gain slider to adjust by 1 dB per notch.

5.2 Pan

The Pan slider positions the signal in the stereo field, from L50 (hard left) through centre to R50 (hard right). FusionDesk uses a constant-power (sin/cos) panning law — the perceived loudness stays consistent as a signal moves across the stereo field.

At centre, both channels receive −3 dB. This is correct constant-power behaviour and matches professional console convention.

5.3 Width & Mono Mode

The Width slider controls stereo width, from 0% (mono sum) to 100% (full stereo). Narrowing the width blends the left and right channels together — useful when a stereo source sounds too wide or when you need a tighter, more focused image.

Mono mode locks the channel to true mono. Right-click the channel name and select Mono to toggle it. In mono mode:

• Input strips: a single ASIO channel is copied to both L and R, rather than using a stereo pair
• Output strips: L and R are summed to (L+R)/2 before the VST chain
• The Width slider is hidden and the right meter is suppressed

Mono mode state is saved with the session.

To set multiple channels at once, use the main menu Channel Strips section:

• Inputs Mono / Outputs Mono / All Mono — sets all visible input strips, output strips, or both to mono
• Inputs Stereo / Outputs Stereo / All Stereo — returns them to stereo

The Main and Monitor output strips are always stereo and are not affected by these actions.

5.4 The Fader

The fader is the main level control. Range: −∞ (silence) to +6 dB. All faders default to −∞ on first launch to prevent accidental full-volume output.

On input channels, the fader level applies per mix — each mix remembers its own fader position for every input. On output channels, the fader is global and does not change per mix.

5.5 Mute

The Mute button silences the channel. Monitoring is not affected.

6. Mixes — The Core Concept

6.1 What a Mix Is

A Mix is a complete set of routing decisions: which inputs are assigned, at what fader levels, panned where, and sent to which output strip. FusionDesk supports up to 32 simultaneous mixes.

Think of it as having 32 independent mixers running at once, sharing the same inputs but each sending a different blend to a different destination.

Practical examples:

Mix 1 → Main speakers (full band mix)

Mix 2 → Vocalist's headphones (more vocals, less drums)

Mix 3 → Drummer's wedge (more click, less everything else)

Mix 4 → Livestream encoder (mix-minus, no talkback)

Per-channel controls — Gain, Pan, Width — are global and do not change per mix. Only the fader level is mix-specific.

6.2 Selecting and Switching Mixes

Click a Mix button at the bottom of any input strip to select that mix. The active mix is shown with an orange background and black text; inactive mixes show as dark grey.

Switch mixes with the < / > keyboard shortcuts, or click any Mix button directly.

6.3 The Routing Indicator (Green Dots)

When an input strip is selected, a small green dot appears on each Mix button where that input has a non-zero fader level. This gives instant visual feedback showing which mixes the selected input feeds into.

Conversely, selecting a mix button shows which inputs are routed to it — any channel with a fader above −∞ in that mix will show activity.

6.4 Copying, Pasting, and Clearing Mixes

Right-click any Mix button for a context menu:

• Copy — copies all fader levels for the current mix
• Paste from Mix N — pastes a previously copied mix state
• Clear Mix — resets all input faders in this mix to −∞

6.5 Linking Mixes (Link Groups)

To link mixes: right-click a Mix button → Copy, then right-click the mix you want to link to → Link to Mix N. When mixes are linked, a fader move on any member of the group is reflected on all others in real time — useful for keeping a headphone mix locked to your main output level.

To remove a link: right-click → Remove Link.

6.6 Matrix View

Menu → Matrix View... (or press Alt+M) opens the routing matrix: a grid with every channel down one side and every mix across the other. Each square is the crosspoint where one meets the other — a green square showing its gain in dB means that channel feeds that mix, and a dark square means it doesn't. Esc closes the window.

The channel-strip view answers "what is this channel doing?" The matrix answers "what is everything doing?" It is the router's view of the same routing the main surface already edits, and the two stay in step — a change in either shows up in the other immediately.

Editing crosspoints — four ways:

Dragging takes its on/off state from the first square you touch and applies it to every square you cross, so a drag is always all-on or all-off — and dragging back over a square does not undo it. Scrolling only trims squares that are already on; scrolling anywhere else scrolls the grid.

The Outputs master row. Along the bottom of the grid, an Outputs row gives each output its own cell showing that output's fader level in dB. These cells work exactly like the crosspoints above them — click to switch the output on or off, scroll to trim its level (±1 dB a notch, ±0.5 dB with Alt), and double-click to type an exact value — but they act on the output's own fader, so setting a level here moves the matching output strip and vice versa. It lets you set your output levels from the same grid you route in.

Transpose flips the axes, putting channels across the top instead of down the side. This suits a wide screen with many channels and few mixes. The corner of the grid always names both axes, so you can tell at a glance which way round you currently are. The Outputs master cells move with their outputs — along the bottom normally, and down the right-hand edge when transposed — so they always sit beside the outputs they control.

The window sizes itself to the grid when you open it and whenever you change the channel count, and it holds a sensible minimum so it can't shrink into a sliver. Two buttons on the bar fine-tune it: Resize snaps the window to fit the grid exactly — the last channel against the right edge, the last mix just above the buttons — and Centre returns the window to the middle of its monitor if it drifts off (for example after dragging it between screens). Neither sizes the window beyond your screen, so with many channels you may still need to scroll. The window also remembers the size and position you leave it at.

When the grid is larger than the window it scrolls, and the name bands stay put as you go: scroll down and the headings across the top remain visible, scroll sideways and the names down the side remain visible. Each name stays locked to its own row or column, so you always know which crosspoint you are looking at.

Linked mixes behave here exactly as they do on the faders — editing a crosspoint on a linked mix moves its linked partners at the same time, whether or not that mix is the one currently selected.

Tick Reopen on startup (bottom-left of the window) to have the Matrix View come back automatically each time you launch FusionDesk.

7. Metering

7.1 Metering Types

FusionDesk displays three complementary metering types:

• RMS — perceptual loudness. Shown as the main meter bar.
• RMS Peak — the highest RMS loudness level reached. Shown as a blue tick mark on the meter bar and numeric readout below the meter.
• Sample Peak (dBFS) — the highest individual sample value, shown as the orange tick mark on the meter bar. This is the classic sample-domain peak (not inter-sample), kept on the bar as a familiar reference.
• True Peak (dBTP) — the highest level the signal actually reaches, including the peaks that fall between samples. Shown as a numeric readout above the meter. The True Peak number is the one to watch for converter/encoder overs.
• LUFS — Loudness Units Full Scale, per EBU R128. Displayed in the Control Room for the monitor output only.

Hover any meter readout to see a short reminder in the title bar (enable Show Tooltips in the Window menu): the RMS peak number, the True Peak (dBTP) number, the clip indicator, and the orange dBFS-peak tick each name themselves on hover.

The True Peak readout (per ITU-R BS.1770) is important because the real waveform can rise above the highest stored sample — so a signal can read 0 dBFS yet still overshoot and clip a converter or downstream encoder. A brickwall limiter holding 0 dBFS, for example, will typically still show a small positive True Peak (e.g. +0.1 dBTP) because of these inter-sample peaks. For efficiency the inter-sample calculation only runs when a channel is already near full scale (it can make no difference at lower levels), so it adds no measurable load on a quiet mix.

Two separate "over" signals — they mean different things, on purpose:

• The True Peak number turns red at 0 dBTP and above. This is a ceiling warning — it tells you the signal has no inter-sample headroom left for a converter or encoder, which can happen before any sample actually clips. (Streaming targets such as -1 dBTP mean the number is your guide long before a clip.)
• The clip indicator (the light) lights on a sample clip — by default a true 0 dBFS hard over. It deliberately ignores inter-sample peaks, so a limiter sitting cleanly at 0 dBFS reading +0.1 dBTP does not trip the clip light. The trip level is adjustable: see Clip threshold in Section 15.1 if you would rather be warned a few dB early.

So it is normal and correct to see the red True Peak number with a green (un-clipped) light: it means "you're at the true-peak ceiling but you haven't hard-clipped a sample." Hover either readout (with Show Tooltips on) for a one-line reminder of which is which.

7.2 Metering Modes

Open Settings → Metering to choose a ballistic preset:

Click Manage… next to the dropdown to open the Ballistics Manager. Here you can create, edit, copy, and delete your own named presets. Each preset has three parameters:

• Rise (ms) — how quickly the meter responds to a rising signal
• Hold (ms) — how long the meter holds at its peak before falling
• Fall (ms) — how quickly the meter falls after the hold expires
• Instant drop — when ticked, the meter drops instantly to zero after the hold (like Peak dBFS Max Hold)

Click New to create a preset with default values, or Copy from… to start from any built-in or existing custom preset. Custom presets are saved and persist between sessions.

The Peak tick marks option (Settings → Metering) shows the RMS peak and dBFS peak marker bars on the meter. Turn this off for a cleaner meter display. In VU, PPM BBC, PPM EBU, and PPM Broadcast modes, tick marks are always suppressed — the checkbox is greyed out.

In VU and PPM modes, the peak number readouts (RMS Peak and True Peak) use a fixed 1.5-second hold with a slow decay, independent of the meter ballistics. This keeps the numbers readable at slow-moving ballistics without coupling them to the bar's attack/release.

7.3 Signal Presence & Clip Indicator

A small light at the top of each channel shows whether audio is present. It lights green when a signal is detected, yellow while MIDI notes are held, and red when the signal has clipped. The red state stays lit until you reset it.

To reset: with Clip armed in the Control Room panel, click the Reset button next to it to clear all clip indicators at once.

7.4 LUFS Display (Momentary & Short-Term)

The Control Room panel shows LUFS-M (momentary, 400 ms window) and LUFS-S (short-term, 3 s window) for the monitor output. These follow EBU R128 K-weighting and are useful for broadcast, streaming, and mastering compliance work.

Two Max hold buttons sit alongside the LUFS display. Peaks Max holds the highest RMS peak, dBFS Peak, and True Peak values on every meter. LUFS Max holds the highest LUFS-M and LUFS-S values. Both stay held until you deactivate the button.

7.5 The dB Scale

FusionDesk uses a non-linear dB scale across the full meter height. The scale is divided into three regions:

This gives expanded resolution in the working range (+6 to −10 dB) where mixing decisions happen, while keeping the low-level and noise-floor regions readable at any window height. Scale labels are suppressed if they would overlap at reduced window heights.

8. Monitoring & Listen Routing

Monitoring in FusionDesk revolves around one idea: the Monitor channel — what you hear on your headphones or control-room speakers — operates in one of two modes at any moment.

• Mix Mode is the normal state. The Monitor plays its assigned mix, through its own VST chain (e.g. headphone EQ or room correction). This is what you hear while you work.
• Listen Mode is temporary. The moment you engage any PFL, AFL, or Solo button — on any channel — the Monitor stops playing the mix and instead plays only the thing you asked to listen to. Release every listen button and it drops straight back to Mix Mode.

In other words: Listen Mode is automatic. There's no button called "Listen" — pressing PFL/AFL/Solo is how you enter it, and letting go is how you leave. The rest of this chapter explains the three ways to listen (PFL, AFL, Solo) and how to route the listen signal to a separate output.

A green selector tells you which mode you're in. In Mix Mode the Monitor channel's own device selector is green — the mix is going there. The moment you enter Listen Mode the green jumps to the Listen Output selector instead, marking where the listen signal is now being heard. (If no separate listen output is assigned, both are the same device.) Watch the green to see at a glance whether you're hearing the mix or a listen tap.

The one thing to remember: the Monitor shows the full mix until you press a listen button, then it shows only what you're listening to. That switch is Mix ↔︎ Listen Mode — and the green selector follows it.

8.1 Monitor Channel

The Monitor channel represents your control room or headphone output — independent of what goes to the room (Main).

Its signal path is the same in both modes: whatever it is playing — the assigned mix (Mix Mode) or the listen signal (Listen Mode) — passes through the Monitor's VST chain (e.g. headphone EQ, room correction) and monitor fader, then to the output device. Because the listen signal runs through the same chain, your monitoring processing still applies to whatever you're auditioning.

8.2 PFL — Pre-Fader Listen

The PFL button (on every input and output channel) sends the pre-fader, pre-VST signal to the Monitor output. PFL never mutes other channels — it simply adds the pre-fader signal to the listen bus.

PFL is always active regardless of whether the Monitor is in Mix Mode or Listen Mode. It is independent of AFL and Solo.

8.3 AFL Mode

When the Monitor is in AFL mode, the AFL/Solo button on each channel routes the post-VST signal to the Monitor output at a consistent level, independent of the channel's fader position. AFL does not mute other channels — it is a listen-only tap.

8.4 Solo Mode

When the Monitor is in Solo mode, the AFL/Solo button routes the channel's signal to both the Main output and the Monitor output, and mutes all other channels. Solo is the standard "isolate this channel" function familiar from any mixing desk.

Solo is scoped to the active mix bus — other mixes are unaffected.

8.5 The Control Room Mode Switch

The AFL / Solo toggle in the Control Room panel switches all channels' AFL/Solo buttons between the two modes simultaneously. The button label on every channel updates to reflect the current mode ("AFL" or "S").

The Monitor channel's own AFL button is disabled (it cannot AFL into itself).

8.6 Listen Output Selector

By default, PFL/AFL/Solo is heard on the Monitor channel's assigned output. The Listen Output selector in the Control Room panel lets you assign a separate output device for listening — any ASIO pair, Windows Audio device, or Default Output.

When a listen output is assigned, the Monitor mix continues playing on the monitor output uninterrupted while PFL/AFL/Solo plays on the listen output. The active device selector is shown in green.

8.7 Monitor PFL — A/B VST Bypass

Pressing PFL on the Monitor channel itself bypasses the monitor VST chain (headphone EQ, room correction, limiter) and passes the dry mix directly to the monitor output. This is intentional — use it to A/B quickly between your processed and unprocessed monitoring without changing any settings.

This does not trigger Listen Mode and does not affect other channels.

9. VST3 Plugin Processing

9.1 Plugin Sandboxing — How It Works

Every channel in FusionDesk can host a chain of up to 6 VST3 insert plugins. What makes FusionDesk's approach different is sandboxing: each plugin runs in its own isolated background process, completely separate from the mixer.

If a plugin fails, FusionDesk keeps running. Because the plugin is in a separate process, its failure can't take the mixer down: the rest of your session, every other plugin, and every other channel carry on. The faulty slot is bypassed automatically — the dry signal passes through in its place — and marked with a yellow indicator. There's no emergency restart and nothing is lost, and the slot can be reloaded with a single click.

Whether you hear anything depends on how the plugin fails:

• A plugin that crashes (the common case — an access violation while processing) is caught in the same audio block. That block passes through dry and the slot is faulted instantly, so audio continues without a drop-out — the plugin simply stops contributing. Most failures are of this kind.
• A plugin that hangs or aborts (wedged and never returning, or an outright process exit) can't be caught mid-block, so FusionDesk falls back to a watchdog: after roughly half a second of dry pass-through on that one channel, the slot is faulted and audio recovers. Here you may hear a brief interruption — on that channel only, never the whole mixer.

9.2 Loading a Plugin

Click any channel name to show its VST slots in the Control Room — or, for the focused channel, press Shift (input) or Ctrl (output). Click any empty slot to open the Plugin Browser. Double-click a plugin in the browser or press Enter to load it. New plugins always load bypassed — click the dot on the left of the slot to enable them.

9.3 The Plugin Browser

The plugin browser has four columns:

• Favourites — a curated shortlist; drag plugins here from any column
• Effects Plugins — every effect found across the scan paths (and anything not classed as an instrument); use the search box to filter
• MIDI Instruments — plugins set up to be played with MIDI. FusionDesk makes a best guess when it first sees a plugin; drag any plugin in or out to suit how it is actually used
• Ignore — plugins hidden from the picker

Drag between columns to organise — the arrangement is remembered. Effects Plugins is the catch-all: anything not marked as an instrument or ignored stays there, so nothing is ever lost. The browser scans on open and updates live — opening it is fast (~100 ms).

Add extra scan paths in Settings → VST Scan Paths if your plugins are installed in non-standard locations. The browser rescans automatically each time it opens.

9.4 Slot UI

Each slot shows:

Drag a slot name to reorder plugins within the chain.

9.5 Bypass / Fault / Reload

The slot dot colour tells you the plugin's state at a glance: green = active, yellow = bypassed, red = faulted.

• Bypass (yellow dot): click the dot or use the slot menu. Bypassed plugins are skipped entirely — dry audio passes through.
• Fault (red dot): a plugin has crashed. Audio falls back to dry pass-through. Use the slot menu → Reload to restart it.
• Reload: for a running plugin, reload restarts it in place. For a faulted plugin, reload starts a fresh isolated host for it.
• Unload: removes the plugin from the slot. If the same plugin is used elsewhere, those slots keep running.

Faulted plugin state is not saved at session exit — this prevents a crash-loop on next startup.

MIDI-effect plugins in the chain. If a channel has a MIDI instrument, you can place MIDI-effect plugins — a chord generator, arpeggiator, or MIDI delay — earlier in the same chain, ahead of the instrument. The MIDI flows through them in slot order, each one transforming the stream before it reaches the next: play one key into a Chord → Arpeggiator → Instrument chain and you hear a chord, arpeggiated at the strip's tempo. Reorder the slots (drag a slot name) to change the result — an arpeggiator before a chord generator sounds different from the reverse. Note timing is preserved sample-accurately through the chain, so a fast arpeggiator stays tight regardless of buffer size.

9.6 Editor Windows — Position & Multi-Monitor

Plugin editor windows float freely and stay above the FusionDesk window without being system-wide topmost (they don't cover unrelated apps).

Window positions are saved per slot in physical pixels, DPI-aware, and restored on the next session even across monitor changes. To reset a window's position to the default (centred on the strip area), uncheck the Save window position in the slot menu (clears the stored position).

Editor windows can be dragged across monitors freely. A brief size adjustment may occur when crossing a DPI boundary — this is a framework limitation and settles immediately.

9.7 Plugin State Persistence

VST plugin settings are saved automatically when FusionDesk exits and when a plugin is reloaded. Your settings are restored on the next session automatically — no extra steps needed.

9.8 Copying a Plugin Chain

Right-click the channel name or device label above the VST slots in the Control Room to access the chain clipboard:

The clipboard is in-session only — it is cleared when FusionDesk exits.

9.9 Native EQ

A built-in parametric equaliser — up to 24 bands, edited directly on an interactive response curve. It sits alongside the other native processors in the picker and adds no plugin-host overhead.

To use it: click an empty slot → Plugin Browser → Native Plugins section at the top → EQ. It loads bypassed with one band ready to shape; click the dot to enable it.

Working on the curve (the fastest way to use the EQ):

• Drag a point to set its frequency (left–right) and gain (up–down). Cut bands move left–right only.
• Double-click empty space to add a new band at that frequency and gain.
• Double-click a point to remove that band.
• Right-click a point to bypass that band — the point turns red and grows, and the band drops out of both the curve and the sound without losing its settings. Right-click again to bring it back. You can bypass as many bands as you like at once, which makes it easy to A/B a single move against the rest of the curve.
• Scroll wheel over a point (or anywhere, for the selected band) to adjust its Q — hold Alt for fine steps.
• The selected point shows a yellow ring; its settings appear in the control bar below.

Each band draws its own faint curve in its own colour behind the combined response, so you can see what every band contributes. A live spectrum analyser also runs behind the curve, showing the frequency content of the signal as you shape it. It uses the same scale as the meter to the right (a faint red line marks 0 dBFS). Note that, like any analyser, a strong low-frequency tone can read a few dB above 0 dBFS even when the channel isn't clipping — that reflects the energy in that one band, not the overall level.

Two toggle buttons next to the Output control adjust the analyser display:

• Peak — holds a line at each frequency's recent maximum, easing back down slowly. This catches short peaks that the moving spectrum would otherwise flash past, so you can see where a signal is really reaching.
• Freeze — holds the whole display still so you can study a moment in detail. Switch it off to resume the live display. (With Peak also on, the held line freezes too.)

Band types (choose from the Type dropdown in the control bar):

• Bell — boost or cut around a centre frequency; Q sets the width.
• Low Shelf / High Shelf — boost or cut everything below / above the frequency.
• Low Cut / High Cut — remove everything below / above the frequency. Instead of gain, these use a Slope (6, 12, 24 or 48 dB per octave) — gentler to steeper.
• All Pass — passes all frequencies at full level but shifts their phase around the frequency; no change to the tone, used for phase alignment.
• Band Pass — a single resonant band that passes a range around the frequency and rolls off either side; Q sets how narrow. With a very low Q it becomes a broad tonal dome.
• Tilt Shelf — an S-curve that lifts one end of the spectrum and lowers the other, pivoting around the frequency.
• Flat Tilt — a straight-line tilt across the whole range: a constant slope that pivots at the Freq point (which stays at 0 dB) while everything below tilts one way and everything above the other. Gain sets how much it tilts; the curve eases off only at the very edges of the range. (Slope is not used for this type.)

Parameters (all settings persist across sessions):

• Freq: 10 Hz – 20 kHz per band.
• Gain: −30 to +30 dB per band (Bell and Shelf types).
• Q: 0.01 – 20 (Bell and Shelf types) — higher is narrower.
• Slope: 6 / 12 / 24 / 48 dB per octave (Cut types).
• Output: −12 to +12 dB — a global trim after all bands, for matching levels when the EQ adds or removes energy.

Use Presets to load factory starting points, save user curves, or open the preset folder.

The sliders in the control bar mirror the curve: drag, scroll (Alt for fine), or double-click a number to type an exact value. The Freq value also accepts musical note names — type A4, C#3 or Gb5 and it sets the matching frequency (A4 = 440 Hz). Alt + click any slider to reset it to its default; to reset the whole EQ, right-click its slot and choose Reset to Default Settings.

The window is fully resizable — drag any edge to make the curve as large as you like, and the size you choose is remembered for the next time you open it. Like the other native dialogs it stays open while you work in the mixer, and Save window position (right-click the slot) reopens it where you left it.

9.10 Native Compressor

FusionDesk also includes a VCA-style compressor as a built-in plugin — no third-party installation required. It sits alongside the other native processors in the picker.

To use it: click an empty slot → Plugin Browser → Native Plugins section at the top → Compressor. It loads bypassed; click the dot to enable it.

Detection is stereo-linked — both channels receive the same gain reduction so the stereo image stays intact. The detector follows the louder of the two channels — unless you set a sidechain (see below), in which case it follows that channel instead.

Parameters (all settings persist across sessions):

• Sidechain (top of the dialog, default None) — the channel the compressor listens to when deciding how much to turn down. Leave it on None and the compressor reacts to its own signal, as usual. Choose any Input or Output channel and it reacts to that channel instead, while still processing its own audio — the classic "ducking" setup, e.g. a kick keying the bass, or music ducking under a voice. Sources are grouped into Inputs and Outputs submenus (so two channels sharing a name are never confused), a compressor can't pick its own channel, and the plugin slot shows the wiring as "Compressor - Kick". When you key an output compressor from an input, or vice versa, detection is current-block; keying an input compressor from an output is one buffer late (about 3 ms at 48 kHz / 128 samples) — inaudible for ducking. The setting saves with your mix and travels with copy/paste; if the source channel later goes out of range (you reduce the channel count), the sidechain quietly reverts to None.
• Threshold: −60 to +6 dBFS (default −12 dBFS) — the level above which gain reduction begins. The range extends above 0 dBFS so you can keep the compressor fully "out" on hot sources that already peak above full scale. The defaults give gentle, audible compression on typical program material on first open.
• Ratio: 1:1 to 20:1 (default 2:1) — how much above-threshold signal is reduced
• Attack: 0.1 to 100 ms (default 10 ms) — how quickly gain reduction engages
• Release: 10 to 2000 ms (default 120 ms) — how quickly gain reduction recovers
• Knee: 0 to 24 dB (default 6 dB) — soft-knee region centred on the threshold; higher values smooth the onset
• Makeup: −24 to +24 dB (default 0 dB) — output gain applied after compression to compensate for level loss

Editing parameters:

• Drag the slider thumb up or down. Clicking the track does not jump the value — you move a control only by dragging its thumb (the same as the main mixer faders).
• Scroll wheel over a slider steps in coarse units: Threshold/Knee/Makeup 1 dB, Ratio 1:1, Attack 5 ms, Release 20 ms. Hold Alt for finer steps: 0.5 dB, 0.1:1, 1 ms, 5 ms respectively.
• Double-click the number to type an exact value; Return confirms, Escape cancels.
• Alt + click the slider to return that parameter to its default.
• To reset every parameter at once, right-click the Compressor in its slot and choose Reset to Default Settings — if the dialog is open, the sliders update live.
• Close closes the dialog; clicking the compressor slot again also closes it.

The dialog stays open when you click elsewhere in the mixer — close it explicitly when you're done. Right-click a Compressor slot and tick Save window position to have the dialog reopen where you left it.

Metering: the dialog includes a full strip-style meter showing the post-compressor output level — green/yellow/orange/red bands, RMS level + sample-peak indicators, full −70 to +6 dB scale — flanked by orange gain-reduction bars anchored at the Threshold. The meter follows your global meter ballistics, peak-hold mode, and RMS reference from Settings. The strip itself also shows orange GR bars flanking its L/R meters.

History graph: a 10-second scrolling waveform behind the controls shows the post-compressor peak level (white line) and gain reduction over time (red line), with a faint horizontal line marking the current threshold. The graph's vertical scale matches the meter alongside it, so the threshold line sits at the same height as the threshold value on the meter — signal pushing above the line is being compressed. Useful for seeing at a glance whether the compressor is catching the transients you expect, whether the ratio is doing what you want, and whether attack/release shapes match the program material. The graph freezes when the compressor is bypassed.

Combined GR display: when both a Compressor and Limiter are active on the same channel, the strip's GR bar shows the combined reduction across all active processors, anchored at the highest threshold. Bypassed processors are skipped.

Because the compressor occupies a normal insert slot, you can place it anywhere in the chain — before or after other plugins — and load it on as many or as few channels as you need.

9.11 Native Limiter

FusionDesk includes a brickwall limiter as a built-in plugin — no third-party installation required.

To use it: click an empty slot → Plugin Browser → Native Plugins section at the top → Limiter. It loads bypassed; click the dot to enable it.

Parameters (all settings persist across sessions):

Listed in the order they appear in the dialog:

• Threshold: −40 to +6 dBFS (default 0 dBFS) — the output ceiling. Signal above this is attenuated.
• Lookahead: 0–5 ms (default 0) — non-zero values give true zero-overshoot brickwall limiting at the cost of a small processing delay
• Attack: 0–5 ms (default 0 = instant) — how quickly the gain reduction is applied. See the note below.
• Release: 1–500 ms (default 100 ms)
• Knee: 0–12 dB (default 0 = hard knee) — higher values smooth the onset of gain reduction around the threshold
• Gain: 0 to +30 dB (default 0 dB) — drives more signal into the limiter without raising the output ceiling above the Threshold. Pushing Gain up produces more gain reduction rather than a hotter output.

Attack and Lookahead work together. Attack softens the harshness you get when the limiter clamps hard at low thresholds, by letting the gain ramp down rather than snap to its target. But while the gain is still ramping, peaks pass through underneath it and are caught by the output ceiling instead — which clips them.

Lookahead buys the time that ramp needs. As a rule of thumb, set Lookahead to about twice your Attack and peaks will be under control before they arrive. With Lookahead at 0, any Attack above 0 will let some peaks reach the ceiling — usable as a deliberately softer, less clinical sound, but not transparent limiting. At Attack 0 (the default) the limiter behaves exactly as it always has, and Lookahead alone governs overshoot.

Editing parameters:

• Drag the slider thumb up or down. Clicking the track does not jump the value — you move a control only by dragging its thumb (the same as the main mixer faders).
• Scroll wheel over a slider steps in coarse units: Threshold/Knee/Gain 1 dB, Release 10 ms, Lookahead and Attack 0.5 ms. Hold Alt while scrolling for finer steps: 0.5 dB, 5 ms, 0.1 ms respectively.
• Double-click the number to type an exact value; Return confirms, Escape cancels.
• Alt + click the slider to return that parameter to its default.
• To reset every parameter at once, right-click the Limiter in its slot and choose Reset to Default Settings — if the dialog is open, the sliders update live.
• Close closes the dialog; clicking the limiter slot again also closes it.

The dialog stays open when you click elsewhere in the mixer — close it explicitly when you're done.

Metering: the dialog includes a full strip-style meter showing the post-limiter output level — green/yellow/orange/red bands, RMS level + sample-peak indicators, full −70 to +6 dB scale — flanked by orange gain-reduction bars anchored at the Threshold. The meter follows your global meter ballistics, peak-hold mode, and RMS reference (EBU/SMPTE) from Settings. The strip itself also shows the same orange GR bars flanking its L/R meters.

History graph: a 10-second scrolling waveform behind the controls shows the post-limiter peak level (white line) and gain reduction over time (red line), with a faint horizontal line marking the current threshold. The graph's vertical scale matches the meter alongside it, so the threshold line sits at the same height as the threshold value on the meter. Useful for seeing whether the limiter is just kissing the threshold or hammering it, and whether the release time is producing the recovery shape you want. The graph freezes when the limiter is bypassed so the last trace stays visible alongside the controls.

Because the limiter occupies a normal insert slot, you can place it anywhere in the chain — before or after other plugins — and load it on as many or as few channels as you need.

9.12 Native Tone Generator

A built-in signal generator for line-up, calibration and testing — no third-party plugin required. It sits alongside the other native processors in the picker.

To use it: click an empty slot → Plugin Browser → Native Plugins section at the top → Tone Generator. It loads bypassed; click the dot to enable it.

Unlike the other native processors, the Tone Generator produces a signal rather than processing the incoming one. The Mix control blends the generated tone over the channel's existing audio: at 0% you hear only the channel, at 100% only the tone, and anywhere in between a blend of the two.

Waveform: choose from the dropdown at the top of the dialog, which shows a small drawing of each shape:

• Sine — a pure tone, no harmonics. The standard line-up signal.
• Triangle / Square / Saw Up / Saw Down — harmonically rich waveforms, band-limited so they stay clean (no digital "buzz") even at high frequencies.
• White Noise — equal energy per hertz (bright).
• Pink Noise — equal energy per octave (the standard for system tuning and room measurement).

Parameters (all settings persist across sessions):

• Freq: 20 Hz – 30 kHz (default 1 kHz) — the pitch of the periodic waveforms. Greyed out for White/Pink Noise, which have no single frequency.
• Level: −70 to +6 dBFS (default −30 dBFS) — the output level of the tone.
• Mix: 0 – 100% (default 100%) — how much of the tone is blended over the channel's existing signal.

Editing parameters:

• Drag the slider thumb up or down. Clicking the track does not jump the value. The Freq slider is logarithmic, so equal travel covers equal musical intervals.
• Scroll wheel over a slider steps in coarse units; hold Alt for finer steps.
• Double-click the number to type an exact value; Return confirms, Escape cancels.
• Alt + click the slider to return that parameter to its default.
• To reset every parameter at once, right-click the Tone Generator in its slot and choose Reset to Default Settings.

The Freq field also accepts musical note names such as A4, C#3 or Gb5.

Metering: the dialog includes a strip-style meter showing the generator's output level, so you can confirm a sine sits at your chosen reference (e.g. −18 dBFS) before sending it down the line.

Tip: load the Tone Generator on an output strip with Mix at 100% to send a clean line-up tone to that output, or on an input strip to inject test signal into a mix. Because it loads bypassed, it stays silent until you switch it on.

9.13 Native Alignment Delay

A built-in alignment delay — a pure time offset for lining a channel up against other feeds. It is not an echo or effect delay: there is no feedback and no wet/dry mix, just a clean delay of the signal by a set amount. It sits alongside the other native processors in the picker and adds no plugin-host overhead.

To use it: click an empty slot → Plugin Browser → Native Plugins section at the top → Alignment Delay. It loads bypassed; click the dot to enable it.

What it's for — the classic live-sound timing jobs:

• Delay towers — delay a tower's feed so its sound arrives in step with the main PA reaching the same listeners.
• Sub alignment — time-align subwoofers to the mains through the crossover region, where even a fraction of a millisecond changes how the two sum.
• Lip-sync — delay an audio feed to match delayed video.

Setting the delay (range 0–2000 ms):

• Coarse slider — sets the whole milliseconds (0–2000).
• Fine slider — sets the fraction after the decimal point (0.00–0.99 ms), independently of the Coarse slider.
• The total delay is Coarse + Fine and is shown in the number field centred between the two sliders — double-click it to type an exact value in milliseconds.
• Because the two sliders are independent, you can trim the fine part without disturbing the coarse setting (and vice versa) — no more fighting one linked value.
• The delay is also shown in samples for the current sample rate, above the sliders.

Alt + click either slider to reset to 0 ms, or right-click the slot and choose Reset to Default Settings. The delay value saves with the session and snapshots, and travels when you copy a channel's plugin chain to another channel.

Note: the delay is applied to the nearest whole sample, which is more than precise enough for alignment (one sample is around 20 microseconds at 48 kHz). The samples read-out shows exactly what is applied.

Performance: the plugins on a channel run one after another, so a channel's total processing time is the sum of its chain. If you stack heavy processors and start hearing clicks, see Appendix E: Plugin Performance for why this happens and how to fix it.

10. Latency Compensation

10.1 What It Does

When VST plugins are loaded on a channel, they introduce processing latency. Different plugins on different channels add different amounts of delay, which would cause channels to arrive out of step — causing comb filtering and phase issues when signals are combined.

FusionDesk's latency compensation calculates the highest plugin-chain latency across the channels and adds a pre-delay ring buffer to the lower-latency channels, so they all line up. Input channels are aligned to each other (they sum at the mix bus), and output channels are aligned to each other, so both stages stay phase-coherent.

The System Latency readout at the top of the Control Room panel shows the worst plugin-chain latency across all channels (the amount everything is aligned to), naming the channel — for example, 22.5ms (In 7).

10.2 ASIO-Only Scope

Input compensation applies to ASIO input channels only — Windows Audio (WASAPI) and per-application capture sources are excluded, because their buffering is not deterministic and sample-accurate alignment is not achievable for them. Output compensation applies to the ASIO output channels.

10.3 Enabling / Disabling

Latency compensation is enabled by default. To disable it, open Settings → Behaviour → Latency Compensation and uncheck the toggle. Disabling zeroes all delays instantly without interrupting audio. Re-enabling recalculates and re-applies them.

11. Recording

11.1 Linear Recording

FusionDesk can record any armed input or output strip to a continuous file on disk. Click the Record button in the Control Room panel to start; click Stop to finish. An elapsed time display (HH:MM:SS) shows the current take length.

The Split button (linear mode only) closes the current file and starts a new one immediately — useful for splitting a long session into segments without stopping the recording.

Files are named automatically with a timestamp.

Input tap: post-gain, post-pan, post-width, pre-VST.
Output tap: post-fader, post-VST, pre-mute.

Isolated (ISO) tracks. Every armed channel records to its own file, so arming a set of inputs gives you a separate track per source rather than a single mixed-down file. Because the input tap sits before the plugin chain, those tracks carry the source with the channel's gain, pan and width applied but none of its insert processing — the usual meaning of an ISO in live and broadcast work. That is what makes a recording re-mixable afterwards: you can re-balance, re-process or repair a track later without the original plugin settings baked in.

Arming an output at the same time captures the finished mix as it was heard, since the output tap is after the fader and the plugin chain. Recording inputs and an output together is the standard live setup: isolated sources to work with, plus a reference mix of the show as it went out.

Monitor recording behaviour — When you're in Listen Mode (PFL, AFL, or Solo is active), the Monitor channel's record button behaves differently: it records the PFL/AFL signal you're hearing, including the Monitor channel's fader, pan, and VST processing. In normal Mix Mode, the Monitor records the raw mix as usual.

11.2 Loop Recording

Loop recording maintains a fixed-length RAM ring buffer that continuously overwrites itself. Press Capture at any moment to save the last N seconds to disk and immediately restart the buffer — you never lose what just happened.

Configure the loop length (5 s to 5 min) via the Loop Length selector in the Control Room panel.

Memory note: at 48 kHz stereo 32-bit float, a 5-minute loop across 32 armed channels uses approximately 1.2 GB. Ensure sufficient free RAM when using long loop lengths with many armed channels.

11.3 File Formats

For archival quality, WAV 24-bit or FLAC 24-bit are recommended. MP3 is suitable for distribution-quality captures where file size matters.

11.4 Arming Channels

Click the Record button on an individual channel to arm it for recording. Only armed channels are written to disk. The Arm button in the Control Room panel gates this — with the Record Arm button active, clicking a channel's Rec button arms it; with it inactive, channel Rec buttons have no effect.

11.5 Recording Folder

The recording folder defaults to a Recordings folder inside FusionDesk's app-data directory (%APPDATA%\FusionDesk\Recordings). To change it, open Settings → Recording → Recording Folder and click Browse. New recordings go to the new location immediately.

11.6 Recording Info Window

The Info button in the Control Room's recording section opens a small Recording Info window showing free disk space and how long you can keep recording at the current settings. It stays above the main window so you can keep an eye on it while you work the mixer during a take, and it reopens in the same position and size each time — including across restarts and on whichever monitor you left it. Click Info again to close it.

12. Snapshots

12.1 Saving a Snapshot

A Snapshot captures the complete mixer state: every fader, every pan, every routing decision, every plugin and its settings, arm and exclusive states, visual strip order, mix states, channel names, and colours.

To save: open the hamburger menu (≡) → Snapshot → Save. The default filename includes a timestamp and the active ASIO driver name. Files use the .mix extension.

12.2 Loading a Snapshot

Open the hamburger menu → Snapshot → Load. FusionDesk restores the full state from the file. Restore is performance-optimised and completes in under one second for typical sessions.

12.3 Cross-Driver Restore

If a snapshot was saved with a different ASIO driver than is currently active, FusionDesk detects the mismatch and switches drivers automatically before restoring. This may cause a brief audio interruption while the engine reinitialises.

Device selections in the snapshot are matched against available devices by name — partial name matching handles minor driver version differences.

12.4 Configuring the Snapshots Folder

The snapshot folder can be changed in Settings → Recording → Snapshots Folder. If you change the folder, FusionDesk automatically moves any existing .mix files to the new location — no manual file management required.

12.5 Factory Reset

Menu → Factory Reset returns the mixer to its default state. After a confirmation prompt, it resets every channel, all routing, and all processing (faders, pans, widths, mutes/solos, channel names and colours, mix routing, and every built-in and plugin slot) to a clean install.

It is built on the snapshot mechanism — internally it loads a factory-defaults snapshot — so it takes effect live, with no restart. What it deliberately keeps: your selected audio devices, all preferences (meter mode, recall/dim levels, folders, and so on), the window position and scale, and your PAC (application-capture) favourites. Factory Reset gives you a fresh mix, not a fresh environment — you don't have to set your devices and preferences up again.

13. The Arm & Exclusive System

13.1 Arm Buttons

Five Arm buttons appear in the Control Room panel. Each one gates the corresponding channel button:

All arm buttons start on (armed). With an arm button off, the corresponding channel buttons do nothing — useful for preventing accidental mutes or solos during a performance.

13.2 Exclusive Mode

Four Exclusive buttons sit alongside the Arm buttons:

Exclusive buttons are off by default and can be combined with arm buttons freely. Each clears its button across both the input and output rows.

A channel's PFL and AFL/Solo are always mutually exclusive — turning one on switches the other off — since you can only listen one way at a time. (This is independent of the Exclusive buttons, which govern how a button clears the same button on other channels.) The Monitor channel's own PFL button is exempt, acting as an independent A/B toggle for the monitor VST chain.

13.3 Activity Indicators

Each of the five row labels — Mute, PFL, AFL/Solo, Record, Clip — turns red whenever that function is active on any channel, and returns to grey when none is. This is an at-a-glance check that nothing has been left engaged: a forgotten solo, a channel still muted, a channel armed to record, or a clip that has latched. The check covers every channel, including any hidden on an off-screen bank, so a problem on a channel you can't currently see still shows up. A red Clip label is your cue that something clipped — reset it with the Reset button (see Section 7.3).

The indicators reflect what is actually engaged, regardless of whether the matching Arm button is on — a muted channel shows red on Mute whether or not Mute is armed.

14. Per-Application Audio Capture — Advanced

Requires Windows 11 or Windows Server 2022.

14.1 How It Appears in the Device List

Per-Application Capture (PAC) sources appear at the bottom of the device selector on every input channel. They appear automatically when an application is producing audio — no configuration needed. Simply play audio in an app and it will show up within a second or two.

14.2 Favourites & Memory

When you select a PAC source on a channel, it is added to your favourites automatically. Favourites are remembered between sessions — the source will appear in the selector on your next launch even if the application isn't currently running, ready to be assigned as soon as the app starts producing audio.

Favourites are listed alphabetically in the device selector and in the App Sources submenu of the hamburger menu.

Use Forget in the same submenu to remove a favourite you no longer need.

14.3 Renaming Sources

Some applications report unhelpful or technical process names. Right-click a source in the App Sources submenu → Rename to give it a friendly display name (e.g. "Chrome" instead of "chrome.exe"). FusionDesk persists the rename and applies it everywhere the source appears.

To revert a rename: App Sources → (source) → Revert name.

14.4 Blacklisting

If you want to prevent a source from appearing in device selectors — system sounds, a noisy background app, helper processes — right-click it in App Sources → Blacklist. Blacklisted sources are hidden from all selectors.

To restore a blacklisted source: App Sources → (source) → Un-blacklist.

14.5 Silent Capture Workaround

PAC captures a copy of what the application sends to Windows — the original audio still plays to its normal output device. There is no way to capture silently by muting the app in Windows Volume Mixer (doing so also mutes the capture).

Workaround: Route the application's audio to an output device you aren't using — a monitor headphone jack, a USB mic's playback output, or a virtual audio device (e.g. VB-Audio VBCABLE, free) — and FusionDesk will capture from it while nothing reaches your normal output.

Use Windows Settings → System → Sound → Volume mixer to change the output device for a single app without affecting the system default.

14.6 Electron / WebView2 Apps

Apps that use WebView2 or Electron internally — WhatsApp Desktop, Slack, Discord, and similar — spawn helper processes for audio. FusionDesk walks the process tree to attribute audio to the parent application. This works for most cases but may occasionally show the helper process name rather than the app name. Use Rename to correct it if needed.

14.7 App Restart Detection

If a captured application is closed and reopened, FusionDesk detects the new instance within a few seconds and reconnects automatically. If the app restarts very quickly (for example after a crash recovery), there may be a brief gap in audio.

15. Settings

Open Settings via the hamburger menu (≡) → Settings, or press Alt+S. The window has a list of categories down the left — Metering, Files, Plugins, Live Performance, Behaviour — and shows one category's options at a time; click a category to switch. It opens on whichever category was last used. Where a value is too long to fit its field (a recording or snapshots folder path, or a plugin scan path), hover over it to see the whole thing.

15.1 Metering

Metering preset — choose a ballistic preset from the dropdown. Changes apply live — the meters respond immediately so you can hear and see the difference without closing the dialog. See Section 7.2 for a description of each mode.

RMS reference — controls how RMS levels are displayed:

• EBU / AES17 (default) — full-scale sine wave reads −3 dBFS RMS. This matches every major DAW and broadcast loudness standard (including ITU-R BS.1770 / LUFS).
• SMPTE — full-scale sine wave reads 0 dBFS RMS. Chooses the SMPTE RP 155 convention, which is common in US film/post and Audio Precision analysers configured for SMPTE mode.

The choice is a +3.01 dB display offset only — peak readings and internal DSP are unaffected.

Clip threshold — the level at which a channel's clip indicator lights. The default, 0.0 dBFS (full scale), lights it only on a genuine hard over — the signal has actually run out of headroom. Setting a lower value (−0.5, −1, −2, −3, −6 or −12 dBFS) lights it early, as a headroom warning while there is still room in hand. Broadcast and live work often set this a few dB below zero so a problem is visible before anything is damaged. The setting affects the clip indicator only; meter levels, the True Peak (dBTP) readout and the audio itself are untouched. It applies to every channel — inputs, outputs and the Monitor — and is remembered between sessions.

Peak tick marks — when enabled, adds thin tick marks at the RMS peak and dBFS peak positions on every meter scale. Automatically disabled for VU and PPM modes.

15.2 Files (Recording & Snapshots Paths)

• Recording Folder — where recorded audio files are saved. Click Browse to change.
• Snapshots Folder — where .mix snapshot files are saved. Click Browse to change. Existing snapshots are moved automatically when you change this path.

15.3 Plugins

FusionDesk scans standard VST3 installation paths automatically. Add extra scan paths here if your plugins are installed in non-standard locations. The Plugin Browser rescans automatically each time it opens, so newly installed plugins appear next time you open it.

Reopen plugin windows on startup (default off) — when enabled, any plugin windows that were open when FusionDesk was last closed — VST editors and the built-in EQ, Compressor, Limiter, Tone Generator and Alignment Delay — reappear after the plugins finish loading on the next launch. Windows that were closed at shutdown stay closed.

15.4 Live Performance

The Live Performance section groups together the safety options you'd want locked down during a show.

Live Performance Mode (Settings safety lock) — master checkbox. When on, the five sub-options below become active and locked (greyed out). When off, the sub-options are fully configurable but do nothing — everything behaves normally.

When Live mode is engaged, the title bar indicates the lock is in effect.

If you were in Solo mode when you enable Live Performance Mode with "Lock AFL/Solo to AFL" checked, FusionDesk auto-flips to AFL so the lock is meaningful. The underlying button state is preserved; unchecking Live mode restores the normal AFL/Solo label and colour.

Typical workflow: configure the protections during soundcheck to suit your style, tick Live Performance Mode before the show starts, forget about them. During the show, the only way to change any of those settings is to toggle Live Performance Mode off first.

15.5 Behaviour

15.6 Window Options

Access these via the hamburger menu (≡) → Window:

16. Keyboard Shortcuts

Global

PC Keyboard MIDI

Active on any input channel whose input is set to PC Keyboard MIDI (plays that channel's instrument plugin).

To clear a stuck note, re-select the channel's MIDI device in its input dropdown (sends all-notes-off to that channel).

Strip Focus

Tapping Shift focuses the Input row; tapping Ctrl focuses the Output row. Exactly one strip is highlighted at a time, and each row remembers its last-focused strip (restored on next launch).

Input Strip (Shift)

Output Strip (Ctrl)

Slider Focus

VST Slots

Plugin Picker

Four columns: Favourites | Effects Plugins | MIDI Instruments | Ignore.

dB Value Edit

Double-click the dB readout to edit.

Mouse

17. Troubleshooting

17.1 No Audio / Wrong Device

• Check that the correct ASIO driver is selected in the Control Room panel.
• Verify the channel's device selector shows the intended source or destination.
• Check that the fader is raised above −∞ and the channel is not muted.
• If using "Default Input" or "Default Output", confirm the Windows default device is set correctly in Windows Settings → System → Sound.
• If another application holds exclusive access to the ASIO driver, FusionDesk may fail to open it. Close other ASIO applications and restart FusionDesk.

17.2 Plugin Won't Load / Red Dot

• A red dot indicates the plugin has faulted (crashed during processing). Click the slot's ≡ menu → Reload. (A yellow dot just means you have bypassed the plugin; green means it is active.)
• If the plugin repeatedly faults, try loading it on a different channel to confirm it's a plugin-specific issue.
• Check that the plugin is installed correctly and is a valid 64-bit VST3.
• If the Plugin Browser is empty, check Settings → VST Scan Paths and click Rescan in the browser.

17.3 App Source Doesn't Appear

• Per-Application Audio Capture requires Windows 11. On Windows 10, the App Sources list will be empty.
• Make sure the application is actively producing audio — sources only appear when audio is playing.
• ASIO applications cannot appear in App Sources — this is an architectural limitation of Windows.
• Check whether the source has been blacklisted: hamburger menu → App Sources → look for a Blacklisted section.
• If a helper process name appears instead of the app name, use Rename in App Sources to correct it.

17.4 ASIO Driver Won't Open

• Confirm no other application is holding the driver. Some ASIO drivers are single-client and cannot be shared.
• Try closing DAWs, other mixer software, or any application that uses audio before launching FusionDesk.
• If switching ASIO drivers causes a long delay or freeze, wait up to 15 seconds for the engine to reinitialise before concluding there is a problem.
• After a PC reboot, FusionDesk may not auto-reconnect to the previously-active ASIO driver. Click the ASIO selector and choose Reconnect Current at the top of the dropdown.

17.5 Sending Feedback & Diagnostics

Use the hamburger menu (≡) → Send Feedback to open your email client with a pre-populated bug report. When reporting an issue, include:

• Steps to reproduce
• ASIO driver and buffer size
• Which VST plugins were loaded (if relevant)
• A diagnostics export from Help → Export Diagnostics

17.6 Diagnostics

FusionDesk always keeps a plain-language Flight Recorder log. Menu → Flight Recorder reads it without leaving the app: each entry carries a green, yellow or red dot for its severity, newest first (there is an Oldest First toggle if you prefer to read it in file order), a Session dropdown lets you look back at previous sessions, and a Filter dropdown narrows the view to just errors, just warnings, or both. Tick Live and the view follows events as they happen — handy for watching exactly what FusionDesk does when you change a device or load a plugin. If you scroll up to read something, the view stays put rather than jumping back to the newest line. The window can be resized, maximised, and remembers where you left it. Tick Reopen on startup (bottom-left of the window) to have the Flight Recorder come back automatically each time you launch FusionDesk.

Alternatively, use Help → View Diagnostic Files to open the logs and crash reports on disk, or Help → Export Diagnostics to package them for support.

For intermittent issues, Help → Enable Diagnostics turns on extra detailed logging. A small red dot appears in the title bar while diagnostics are on. While reproducing an issue, press Alt + M to insert a [MARK] line into the log.

17.7 The ASIO selector turned red

If the driver name in the Control Room ASIO selector tints red, the selected ASIO driver failed to open or stopped delivering audio. Most commonly this means the driver DLL is missing, the driver refused to initialise, or the device stalled. Reinstall the driver if necessary, or pick a different driver from the list. The red clears as soon as a driver opens cleanly. Note that many ASIO drivers (including Sound Devices MixPre and Matrix VASIO) silently succeed even when their hardware is off — in those cases the selector stays its normal colour and the only symptom is no audio. See Known Issues for more.

17.8 Choppy or Laggy Interface on a Laptop with Switchable Graphics

If FusionDesk's interface feels sluggish on a laptop's built-in high-resolution (4K) screen — with a faint periodic choppiness to the meters that comes and goes — the cause is usually the graphics chip Windows has assigned to it. Many laptops have two GPUs: a low-power integrated one (e.g. Intel) that drives the built-in panel, and a faster discrete one (e.g. NVIDIA or AMD) for demanding work. The integrated GPU throttles its speed to save power, so the interface can be smooth one moment and choppy the next.

To fix it, tell Windows to run FusionDesk on the discrete GPU:

1. Open Windows Settings → System → Display → Graphics.
2. Add FusionDesk.exe to the list (Browse to its install folder if needed).
3. Select it, click Options, choose High performance, and save.
4. Relaunch FusionDesk.

Setting the Windows power plan to High performance helps further. Alternatively, moving the FusionDesk window to a standard-resolution external monitor is also smooth. This is a Windows graphics-scheduling behaviour on hybrid-graphics laptops, not a fault in FusionDesk.

Appendix A: Audio Quality

FusionDesk is built around a single goal: nothing should be added to the signal that wasn't already there.

The internal mix bus runs in 32-bit floating-point throughout. Once your audio enters FusionDesk, every sum, gain change, pan, width adjustment, and plugin insert happens at full float precision with effectively unlimited internal headroom. Levels well above 0 dBFS pass through cleanly inside the engine — clipping only occurs at the converter, never inside the mix.

There is no internal dither, no truncation, and no hidden gain staging between stages. A signal sent through FusionDesk at unity, with no processing engaged, comes out of the converter bit-identical to the source.

The pan law is constant-power (sin / cos). A signal panned hard left or hard right has the same perceived loudness as the same signal panned to centre. At centre, both legs receive −3 dB — the broadcast and console standard. Width control blends L and R smoothly between full stereo and mono sum without changing perceived level.

Metering is calibrated against EBU / AES17 by default — a full-scale sine wave reads −3 dBFS RMS — matching every modern DAW and the ITU-R BS.1770 / LUFS family of standards. SMPTE RP 155 mode is offered for users who prefer the US film / post convention where a full-scale sine reads 0 dBFS RMS. The choice is a display-only offset; the underlying audio is unaffected.

Measurements

FusionDesk was independently tested against an RME UFX III using a professional measurement rig (2026-04-03).

Frequency response — ruler-flat from 20 Hz to 20 kHz.

THD+N — 0.0009% across the audio band (measured on 80 kHz bandwidth). This is indistinguishable from the UFX III reference and better than most hardware mixers at any price point.

Stereo phase integrity — a null test across 20 Hz–48 kHz shows no phase difference between channels. The slight HF roll-off above ~20 kHz is identical to the UFX III — it is the anti-aliasing filter, not FusionDesk.

In practical terms: FusionDesk's contribution to the sound of your signal chain is the same as a length of wire. What you hear is your interface, your microphones, your speakers, and your plugins. The mixer itself is transparent.

Appendix B: VST Sandboxing

Most mixers and DAWs run their plugins inside the same audio engine that drives the mix. If a plugin misbehaves — an obscure parameter combination, a corrupt preset, a bug triggered only at certain sample rates — it can take the entire host down with it, ending your session and silencing the room.

FusionDesk takes a different approach. Every VST3 plugin loaded in FusionDesk runs in its own isolated audio session, completely separate from the mixer engine. Audio is passed in, processed audio is passed back, and the two sides remain at arm's length at all times. (Isolation is per plugin: if you load the same plugin on several channels, those instances share one isolated session, while every different plugin gets its own — so a crash in one plugin never touches the mixer or any unrelated plugin).

The practical consequence is simple: a plugin failure cannot stop the mixer. The affected slot is bypassed automatically, dry audio passes through in its place, and a yellow indicator on the slot tells you what happened. The rest of the channel — and every other channel — keeps running; there's no emergency restart and nothing is lost.

How much you hear depends on the failure mode. A plugin that crashes (an access violation while processing — by far the most common failure) is caught within the very same audio block: that block goes through dry and the slot is faulted instantly, so audio continues with no drop-out. A plugin that hangs or aborts (wedged and never returning, or an outright process exit) can't be caught mid-block, so a watchdog bounds it instead — after roughly half a second of dry pass-through on that one channel, the slot is faulted and audio recovers. Either way the interruption, if any, is confined to the single channel and the mixer never freezes.

When the time comes to deal with it, click the slot's menu and choose Reload. The plugin is restarted in place, with all its parameters and state intact. If the issue was transient (a bad sample, a one-off bug) it'll come straight back. If the plugin is genuinely broken, you can unload it and carry on.

Plugin parameters and internal state are saved automatically when FusionDesk closes and when a plugin reloads — so even after a crash, your settings come back exactly as they were. Faulted plugins are deliberately excluded from the saved state, so a plugin that crashes on load will not crash-loop your session on the next launch.

The same isolation also means plugin editor windows behave more like external tools than embedded panels. They can be dragged across monitors freely, their positions are remembered per slot, and they don't block the mixer UI even if a plugin's editor temporarily hangs.

This is a meaningful advantage in live and broadcast work — the environments where a mid-show plugin crash is most likely and most costly. For third-party VST3 plugins in FusionDesk, it is a self-healing bypass rather than a session-ending event.

A note on native plugins: the built-in processors that ship inside FusionDesk — the EQ, Compressor, Limiter, Tone Generator and Alignment Delay — are NOT sandboxed. They run inside the mixer engine itself for performance reasons — sandboxing carries a small per-plugin overhead that is worth paying for third-party code, but unnecessary for code that ships as part of FusionDesk and is tested as part of every release. In theory a bug in a native plugin could affect the host; in practice they are deliberately kept simple, well-tested, and fault-guarded against unusual inputs.

Appendix C: Adaptive Resampling

When two audio devices share a session, they need to agree on time. A sample rate of "48 kHz" is never exactly 48,000 samples per second — every clock has a small tolerance, and two devices nominally running at the same rate will always drift apart slightly.

Inside a single ASIO interface this is invisible: every input and output shares the same clock, so no drift can occur. The problem appears when a session combines sources that are clocked independently — a USB interface alongside a built-in soundcard, an ASIO digital mixer running on its own internal clock, a Per-Application Capture source running on the Windows system clock while your ASIO interface runs on its own. These sources all think they are at 48 kHz (or 96 kHz, or whatever you've selected), but their second-by-second rate differs by a few parts per million.

Without correction, the result over time is the familiar slow-crackle: tiny gaps or repeats appear as one device hands samples to the other slightly faster or slower than expected. The crackle may be inaudible at first, then build over minutes, then settle, then return — the classic signature of two clocks beating against each other.

FusionDesk solves this by treating every Windows Audio (WASAPI) device and every Per-Application Capture source as an asynchronous source. Rather than assuming each one is delivering the nominal sample rate, the engine measures what the source is actually producing — sample by sample, second by second — and continuously resamples it into the ASIO engine's clock domain. The ASIO interface remains the master; everything else is gently retimed to match.

The retiming is adaptive. If a source speeds up slightly, FusionDesk's resampler tracks the change. If it slows down, the resampler tracks that too. The correction happens in real time and is never audible as pitch shift or modulation — the adjustments per second are far below the threshold of hearing.

The practical result is that you can mix freely across clock domains without any of the traditional pitfalls. An ASIO digital mixer at 96 kHz, a USB headset at 48 kHz, two browser tabs captured via PAC, a virtual audio device — all of them can be summed into the same mix bus, and they will stay in lock-step indefinitely.

Plugin-latency compensation is applied to ASIO channels (see Section 10), because their alignment is sample-accurate; the input side is what keeps these summed sources in step. Adaptive resampling fills the gap for everything else, ensuring the audio remains continuous and correctly timed even though it cannot be perfectly aligned to the sample.

Appendix D: Errors & Diagnostics

In live and broadcast work, the question is never whether something will go wrong (a flaky plugin, an interface that drops off the bus, a sample-rate change that ripples through the rig) but whether you can keep the show running and, afterwards, explain what happened. FusionDesk treats this as a first-class concern. Four layers work together: the audio keeps flowing when a plugin fails, and a record of events is always being kept so a problem can be understood after the fact, even by someone who isn't a sound engineer.

VST error capture: the show doesn't stop. Every third-party plugin runs in its own isolated session, so a plugin crash can never take the mixer down with it. When a plugin faults, FusionDesk bypasses the slot, passes dry audio through in its place, and flags the slot with a yellow indicator. A plugin that crashes outright is caught in the same audio block, so the mix doesn't even skip; a hang or process exit is bounded by a watchdog to about half a second on that one channel. Either way the rest of the mix keeps running, and you can reload the plugin in place when it's convenient, with its settings intact. This is the front line of error management, and it's covered in full in Appendix B.

The Flight Recorder: an always-on record of what happened. FusionDesk continuously keeps a plain-language log of the significant things it does: devices opening and closing, sample-rate and buffer changes, audio stalls, plugins loading, unloading and crashing, snapshots saved and recalled. Each entry is timestamped and tagged with a severity (information, warning, or error), so a glance down the log tells the story of a session. It runs automatically, with nothing to switch on, and it only writes when something actually happens, so it costs nothing during a quiet show. The last few sessions are kept, so a problem noticed today can still be traced even if it began at yesterday's soundcheck. This is what makes a transient glitch diagnosable after the fact, the kind that leaves no crash and no obvious trace: "it stuttered thirty seconds ago" becomes a line you can actually find.

Crash reporting: a readable account, not just a dump. In the rare event that FusionDesk itself closes unexpectedly, it writes two things side by side: a technical crash dump for the developers, and a companion plain-text report containing the version, the nature of the error, and the last events from the Flight Recorder. That last part matters: a crash arrives with the story that led up to it, rather than as an opaque file that means nothing without specialist tools.

Diagnostics export: one click, one file. When a problem needs to go to support, Help → Export Diagnostics gathers everything relevant (the Flight Recorder logs, any crash reports, the current settings, the list of loaded plugins and their fault history) into a single dated .zip on the Desktop that can be emailed. It works entirely offline (nothing is sent anywhere automatically), and your Windows user name is removed from the text files for privacy. Alongside it, Help → Plugin Health shows which plugins have faulted during the current session and how often, so an unreliable plugin can be identified at a glance; useful both for fixing a rig and for deciding which plugins to trust on the next session.

Taken together, these turn "it didn't work" into something actionable. The audio survives the failure; the failure is recorded; and the record can be read by the person in the room and, if needed, handed to someone who can fix it. Menu → Flight Recorder reads the log in the app, colour-coded by severity, with a live view that follows events as they happen; Help → View Diagnostic Files opens the logs and crash reports on disk if you would rather look at them directly. Tick Reopen on startup (bottom-left of the Flight Recorder window) to have it come back automatically each time you launch FusionDesk — handy if you keep it open as part of your normal setup.

Appendix E: Plugin Performance

When you fill a rig with plugins and start hearing clicks, the instinct is to blame the number of plugins. The real limit is subtler, and understanding it tells you both how to arrange a session and what to look for in a new PC.

Chains are serial; channels are parallel. Within one channel, plugins run in order — the first must finish before the second can start, because each one feeds the next. That chain of processing has to complete inside a single audio buffer. Across channels, though, FusionDesk works on many chains at once; a 32-channel session genuinely spreads its work over all your CPU cores. So the ceiling on a session is not the total number of plugins — it is the deepest single channel. A rig with one plugin on each of 32 channels is comfortable; a rig with six heavy plugins stacked on one channel may click even if every other channel is empty.

The budget. Each audio buffer gives the engine a fixed slice of time: the buffer size divided by the sample rate. At 96 kHz with a 256-sample buffer that is 2.67 milliseconds; at 48 kHz with the same buffer it is 5.3 milliseconds. The deepest chain must finish inside that slice, every time. Miss it once and you get a click. This is why the two levers below work: both give the chain more time.

• Lower the sample rate. Moving from 96 kHz to 48 kHz doubles the time budget and halves the work per second. For most sources 96 kHz buys nothing audible, and it is the single biggest performance lever you have.
• Raise the buffer size. A larger buffer lengthens the budget in exact proportion, at the cost of a little more latency. Moving 256 → 512 doubles the time available.
• Thin the deepest channel. If one channel carries a stack of heavy processors, moving some elsewhere — or removing the ones you don't truly need — attacks the bottleneck directly.

Choosing a PC. Because a single chain is serial, no number of cores can speed it up — one chain runs on one core at a time. The specification that matters for deep chains is therefore single-core performance, not core count. Core count still helps: it lets more channels run in parallel, so a wide session with shallow chains scales beautifully across many cores. But if your sessions lean on deep chains of demanding plugins, prioritise a CPU with strong single-core speed over one that simply has more cores.

This has a real-world consequence that we have measured directly. The identical session — same plugins, same four-deep chain on every channel — ran cleanly on one machine and clicked on another, purely because the second CPU was slower per plugin. Both had ample cores; both parallelised the channels well. The difference was entirely single-core speed on the one serial chain that could not be split. When comparing PCs, the question to ask is not "how many cores" but "how fast is one core."

How to tell if you're at the limit. FusionDesk watches for this. When the deepest chain starts approaching its time budget, a warning is written to the Flight Recorder (Appendix D) noting that a plugin chain is near the audio budget and suggesting a larger buffer or lower sample rate. If you are getting clicks, that log line will tell you whether you have hit this ceiling — and the two levers above are the cure.

Appendix F: CR Readout Calculations

Audio Load

The Audio Load figure in the Control Room shows how much of the audio budget (Appendix E) is being used — that is, how much of the time available to process each buffer is actually spent on audio, expressed as a percentage. It is averaged over a half-second rolling window to smooth out short spikes.

A value of 50% means FusionDesk is using half of the available time for audio processing, leaving the other half as headroom. Values consistently above 80% indicate the system is struggling and audio dropouts become likely. If the Audio Load is high, increasing the buffer size gives the engine more time to process.

The figure changes colour as it climbs, so you can gauge headroom at a glance: it is grey when comfortable, turns yellow above 60%, orange above 70%, and red above 80%.

This figure is not the Windows system CPU usage shown in Task Manager — it measures only the audio processing load, not overall processor usage.

Latency

The simple calculation most people know is:

Latency (ms) = Buffer size ÷ Sample rate × 1000

For example, a 256-sample buffer at 48 kHz gives 256 ÷ 48000 × 1000 = 5.3 ms.

FusionDesk shows a slightly larger figure because it adds the hardware latency reported by the ASIO driver:

Latency In (ms) = (Driver input latency + Buffer size) ÷ Sample rate × 1000 Latency Out (ms) = (Driver output latency + Buffer size) ÷ Sample rate × 1000

The driver latency covers the hardware pipeline — ADC conversion, internal buffering in the interface, DAC conversion — which is real delay that exists on top of the software buffer. Most professional ASIO interfaces report a small but non-zero hardware latency. FusionDesk's figures are therefore more accurate than the buffer-only calculation, though they will look higher than the interface's own control panel shows.

Appendix G: Glossary

ASIO (Audio Stream Input/Output) — A low-latency audio driver standard developed by Steinberg. ASIO drivers bypass the Windows audio stack and communicate directly with audio hardware, delivering the lowest possible latency and the highest channel counts. Requires a compatible audio interface.

AFL (After-Fader Listen) — A monitor mode that taps the signal after the VST chain without muting other channels. Contrast with PFL.

dBFS (decibels Full Scale) — A unit of audio level referenced to digital full scale (0 dBFS = the maximum value a digital system can represent without clipping). All level values in FusionDesk use dBFS.

EBU R128 — A European Broadcasting Union loudness standard. FusionDesk's LUFS metering follows EBU R128 / ITU-R BS.1770 K-weighting.

EQ Curve — The EQ display shows the actual digital filter response at the current sample rate, not an idealised drawing. Bands set very close to Nyquist (half the sample rate) can look slightly asymmetric; this is expected and does not indicate clipping or a drawing bug.

LUFS (Loudness Units Full Scale) — A perceptual loudness measurement. Momentary LUFS (400 ms) tracks short-term dynamics; Short-Term LUFS (3 s) gives a broader view. Used for broadcast and streaming compliance.

Mix — In FusionDesk, a Mix is a complete set of fader levels routing inputs to a specific output bus. Up to 32 mixes can run simultaneously.

PAC (Per-Application Capture) — FusionDesk's term for per-application audio capture via the Windows 11 audio session API. Allows capturing the audio output of individual running applications without virtual cables.

PFL (Pre-Fader Listen) — A monitor mode that taps the signal before the fader. PFL always reflects the source level regardless of the fader position. Pressing PFL on the Monitor channel bypasses its VST chain for A/B comparison.

PPM (Peak Programme Meter) — A meter ballistic standard designed to indicate programme peaks accurately. FusionDesk offers BBC Type I and EBU Type IIa variants.

RMS (Root Mean Square) — A mathematical average of the signal over time, representing perceptual loudness more accurately than peak measurements.

Solo — A monitor mode that routes the selected channel to the main output and mutes all other channels. Contrast with AFL, which routes to the monitor output without muting.

THD+N (Total Harmonic Distortion + Noise) — A measurement of audio quality. Lower is better. 0.0009% is exceptionally low.

VU (Volume Unit) — A classic ballistic standard (IEC 60268-17). Slower than peak metering; reflects perceived average loudness. 0 VU is conventionally calibrated to −18 dBFS in digital systems.

WASAPI (Windows Audio Session API) — The Windows audio API used by FusionDesk for all non-ASIO devices. FusionDesk uses WASAPI shared mode, which allows multiple applications to use the same device simultaneously.

VST3 — Steinberg's third-generation Virtual Studio Technology plugin format. The current standard for professional audio plugins on Windows.

FusionDesk — Mixing, Routing, Processing, Recording
Windows 10 (2004+) / Windows 11 x64
Per-Application Audio Capture requires Windows 11 or Windows Server 2022.