patchd
Guides

Setup guide

Using VST plugins on your microphone, and what patchd does instead

Yes, you can run studio plugins on a live microphone, and this guide explains exactly how people do it. It also says the important part first: patchd does not host VST plugins. Its rack is native. Here is the whole landscape, honestly, so you can pick the trade you actually want.

The honest answer, first

patchd does not host VST plugins. The effects rack on every patchd strip is built from native nodes, processing written into the engine itself, and there is no plugin slot, no VST scanner, no bridge to third-party code. If your voice chain depends on one specific plugin, patchd cannot load it, and nothing below will pretend otherwise.

What this guide can do is more useful than a workaround: explain how people actually get VST plugins onto a live microphone, what that path genuinely costs, and what patchd builds in instead. Both routes are legitimate. They are just different trades, and you should make yours on purpose.

How a plugin gets onto a live mic at all

VST, Virtual Studio Technology, is a packaging standard: a plugin is a piece of audio processing, an EQ, a compressor, a reverb, an amp sim, wrapped in a format any compatible host can load. The key word is host. A plugin cannot run on its own; it has no access to your audio devices and no clock of its own. Something else has to feed it buffers of audio and collect the result. Inside a recording project that something is your DAW, which is why plugins feel effortless there.

A live microphone is the hard case, because the apps you talk through, chat apps, games, meeting software, are not plugin hosts and never will be. They do exactly one audio thing: show you a list of microphones and let you pick one. So the entire problem of VST on a live mic reduces to a trick: make the processed signal show up in that list.

The standard recipe uses two pieces. A plugin-host application reads your real microphone and runs your plugin chain in real time, and a virtual audio device catches the host's output and presents it back to Windows as a microphone. Every other app then selects the virtual device as its mic and receives your processed voice. It works, and it is genuinely powerful: the entire third-party plugin ecosystem, on your voice, live.

What the hosted path costs

Start with the cost you can compute. Every handoff in the hosted relay is a buffer somebody waits for, and the arithmetic is one line: latency_ms = samples / 48000 * 1000. A 256-sample hop at 48 kHz is 5.3 ms. The mic capture buffer is one hop, the host's own processing block is another, and the virtual device that carries the result out is a third. Plugins then add their own reported latency on top of the plumbing: a lookahead limiter needs to see peaks coming, so it buys its catch with a delay, 96 samples is exactly 2 ms, and mastering-grade designs spend 1 to 5 ms. A linear-phase EQ can spend far more.

The hosted billcumulative wait, 48 kHz
51015ms+5.35.3mic capture256 smp+5.310.7host block256 smp+2.012.7lookahead96 smp+5.318.0 msvirtual device256 smppatchd bus path · 10.7 mslatency_ms = samples / 48000 * 1000
Every hop is a buffer you wait for. Two 256-sample hops into the hosted relay, you have already spent the 10.7 ms patchd quotes for its entire bus path, before a single plugin adds its lookahead.

Latency is the predictable cost. The fragile part is that the host is one more real-time application with your microphone's fate attached to it. When it crashes, or hangs while scanning a newly installed plugin, or loses a fight over the device, every program pointed at the virtual device hears silence at the same moment, usually mid-call. And the wiring is per app: each program has to be pointed at the virtual device by hand, and a Windows update, a replugged USB header, or a helpful default-device change can quietly point an app back at the raw mic. Nothing errors. You just sound unprocessed until someone tells you.

What patchd builds in instead

The chain, already racked. High-pass, noise suppression, gate, compressor, in true processing order on the mic strip, with no host and no scanner between them.

patchd's answer is to ship the chain, not the socket. Every strip carries an effects rack of native nodes, added from the + Add Node menu and reordered by dragging, and the free tier includes 13 practical DSP nodes on every strip: gain, high-pass, low-pass, a parametric EQ, noise suppression, a gate, a compressor, a de-esser, upward and downward expanders, a transient shaper, a limiter, and a saturator. That list is not a random sample. It is the chain people are usually chasing when they go looking for mic plugins, and it racks in the canonical order: high-pass first, then noise suppression, then the gate, then the compressor.

Native also changes the design brief. A hosted mastering limiter buys perfect peak catching with 1 to 5 ms of lookahead delay, which is a fine trade inside a DAW where delay costs nothing. patchd's limiter is deliberately zero-lookahead: slightly less perfect catching, in exchange for a node that is safe in a live monitoring path. The nodes were designed for the live-mic job, not borrowed from the studio one.

The outside of the trick stays familiar. A strip routes to a bus, and the bus shows up in other apps' device lists as a microphone named Patchd Bus 1, the same shows-up-as-a-mic idea the hosted recipe uses, except the mic, the processing, and the endpoint live in one engine instead of three cooperating programs. The free mixer guide walks the whole board.

One engine, one measured number

Because there is no inter-app relay, there is exactly one path to account for, and patchd measures it rather than estimating it. The whole trip is about 10.7 ms through the bus at a 512 buffer, and smaller ASIO buffers take it lower. The buffer itself is set in your audio interface's own control panel, and the master clock readout at the top of the app shows the locked source, the sample rate, and the live measured latency of the running engine.

The receipt, live. One readout for the one path: the locked clock source, the rate, and the measured latency of the entire chain, with no second app around to add a hop to it.

So which path should you take?

Say the trade plainly, in both directions. A native rack cannot load third-party plugins; that is the deal, and no honest page skips it. If your sound depends on one specific plugin, a particular character compressor, an amp sim, a processor your workplace mandates, the hosted recipe above is the only route that gets you there. Take it with your eyes open: budget the stacked milliseconds, treat the host as critical software, and re-check your device wiring after every update.

If what you actually wanted was your mic, cleaned up and consistent, in every app, then the plugin hunt was never the goal. It was the workaround. patchd builds that chain into every strip, free, with the routing to carry it anywhere. patchd is in development now. Join the waitlist and we will tell you the moment it is ready to install.

Stop fighting your audio.

patchd is the Windows audio mixer your setup deserves. Join the waitlist to be the first to know when it ships.