Mic processing
How to add effects to your voice in real time
Real time means the effects run live, between your microphone and every app that hears you. This guide builds that path: mic on a strip, nodes from a palette, the order that makes them work, and the handoff that puts the processed voice into your calls and streams.
Adding an effect to your voice in real time is different from fixing a recording afterward: the processing sits in the signal path itself, so your voice passes through the chain while you speak and arrives already processed everywhere at once, the call, the stream, the recording. Nothing is repaired in post, because there is no post. This guide sets that up in patchd, a free virtual audio mixer for Windows, in five moves: put the mic on a strip, add effects to its rack, put them in the right order, monitor the result, and hand the processed voice to the apps that need it. If the phrase DSP effect is new, that explainer covers what these nodes actually do to the signal.
Everything starts at the strip
Add your microphone as an input in patchd and it gets its own strip: a fader for its level, a live meter, M, S, and FX controls, and colored routing pills that decide where the sound goes. The FX control is the door. Click it and the strip's DSP window opens on the effects rack, a vertical list of nodes the signal runs through top to bottom. Whatever this rack holds is what every destination receives. There is no separate set of effects for the call and another for the recording; you build the chain once, on the strip, and every place you route that strip hears the same processed voice.
Add a node from the palette
Every effect enters the chain the same way, through the + Add Node button at the bottom of the rack:
- Click FX on the mic strip to open its rack.
- Click + Add Node. The palette opens with the whole catalog grouped by category: Utility, Dynamics, EQ & Filter, and on down through the creative sections.
- Scroll the catalog, or type in the search field to filter it. The search matches names, descriptions, and aliases, so typing "telephone" finds the Comms node and "comp" narrows the entire catalog to a single card.
- Click the card. The node drops into the rack and is live on your voice immediately; dragging its grip reorders it whenever you like.
What is free and what is Studio
The palette is also the honest version of the pricing page. Everything a spoken voice needs to sound clean and consistent sits in the free sections: thirteen practical nodes covering gain, a gate, a compressor, a limiter, a de-esser, downward and upward expanders, a transient shaper, EQ, high-pass and low-pass filters, noise suppression, and a saturator. The 22 creative voice-fx nodes, reverb, autotune, vocoder, pitch and formant shifters, Monster, Comms and the rest of the character shelf, carry the violet Studio badge. In other words, the practical chain this guide builds costs nothing; the paid line is the fun stuff.
Put the chain in the order the signal wants
The rack runs top to bottom and every node feeds the next, so order is not a style preference; it changes what each node hears. The rule that survives every setup is cleanup first, character later. A high-pass or EQ goes first, because rumble is energy that trips detectors even when you cannot hear it as tone. Noise suppression comes next to strip the steady floor, and the gate sits after it, where the floor is already low, so its threshold can sit far below your voice.
The compressor goes after the gate, never before it: a compressor with makeup gain raises everything, including the room, and a raised floor can climb straight past a gate threshold that used to work. Only after all of that do the creative nodes get their turn, a saturator for warmth, or any of the Studio voice fx. The full reasoning, with the numbers that prove it, is in what order DSP effects should go in.
Hear the result while you tune it
You cannot set a threshold you cannot hear. Light the Playback pill on the mic strip and the strip plays into your own headphones after the rack, so every knob you turn is audible as you turn it. Because the chain is running live, the delay between speaking and hearing yourself is the number that decides whether monitoring feels natural: patchd runs about 10.7 ms through the bus at a 512 buffer, and smaller ASIO buffers take it lower. Roughly 10 ms and under reads as immediate to most people.
If hearing yourself echo slightly behind your own voice bothers you, or you want the buffer math behind that figure, how to monitor your mic with low latency walks the whole path.
Hand the processed voice to your apps
When the chain sounds right, light a bus pill on the mic strip, Bus 1 say. A bus is a virtual output that every other app on your PC sees as a microphone: open Discord, Zoom, or OBS, go to its audio settings, and select Patchd Bus 1 as the input device. The app has no idea a mixer is involved. It receives your processed voice as an ordinary mic signal, gate, compressor, saturator and all, which is the entire point of doing this in real time. The step-by-step with the actual device pickers is in use any app as your microphone, and how to route audio per app on Windows shows how the strips and buses fit together across a whole setup.
Persona is the final stage
If the effect you want is a different voice entirely, that is Persona, patchd's AI voice conversion, and it runs after the whole rack as the pinned final stage, which is why it sits last in the rack figure above. The order logic is the same one more time: clean first, transform second, so the character is built from your words rather than your fan noise. Honesty note: voice conversion is much heavier than the DSP nodes and adds around 350 ms, so it is live but noticeably behind the practical chain. The free tier includes a rotating taster of six voices, and the full library is Studio; the difference between the two kinds of voice changing is covered in DSP vs AI voice changing.
Everything else on this page, the strip, the palette, the thirteen practical nodes, the monitoring, and the bus handoff, is free on every channel. patchd is in development now. Join the waitlist and we will tell you the moment it is ready to install.