What is actually on a radio voice
The sound people call radio voice is mostly processing, and the processing is not exotic. Six effects do the work, and the order they run in is as much of the result as their settings.
- 1High Passcuts the rumble under your voicefirst, so nothing downstream reacts to it
- 2Expanderturns the quiet gaps down furtherbefore the compressor, never after
- 3De-Essertames the hiss on an Swhile the S is still the loudest thing
- 4EQlifts around 3.5 kHz for clarityshapes the tone the compressor will judge
- 5Compressornarrows loud to quiet, then lifts it allafter the noise is already gone
- 6Limitera ceiling a laugh cannot passlast, so nothing can raise the signal after it
1. A high-pass filter
A high-pass filter turns down everything below a point you choose, and around 100 Hz is the usual place for a voice. Almost nothing you say lives down there. What does live there is rumble: a desk being knocked, footsteps through the floor, air conditioning, the low hum of a room.
In a quiet room you will barely hear the difference when you switch it on, which is why people skip it. The reason it goes first is not how it sounds on its own. Rumble is energy, and every effect after it that reacts to level would otherwise be reacting to that energy.
2. A downward expander
An expander is the gentle relative of a gate. When your voice drops below a level you set, it turns the sound down further instead of cutting it off. In the gaps between your words, your fan, your keyboard and your breathing all get quieter, and the transition is soft enough that nobody hears a switch.
It sits before the compressor, and that order is the one worth defending. A compressor raises everything after it has evened things out, so a compressor in front of an expander spends its gain raising the room noise between your words, and the expander then has a louder problem to deal with. Put the expander first and the compressor never meets that noise.
3. A de-esser
Say an S and your mouth makes a short, sharp burst of high-frequency hiss. Close to a microphone that burst can come out louder than the words around it, which is the sound people describe as harsh or sharp.
A de-esser turns down only that hiss and only while it is happening. It is not a permanent cut at those frequencies, which would make every word dull; it is a compressor whose detector listens to the S band alone and acts for the few milliseconds the S lasts. The difference between those two approaches is covered in de-esser vs EQ for sibilance.
4. An EQ, lifting the clarity range
Add an EQ with a modest boost around 3.5 kHz. That is the range where the consonants live, the part of your voice that decides whether a listener catches the word or guesses it. Lifting it makes you easier to understand without making you louder.
Keep it modest. Too much of that range and you cross from clear into harsh, and you will be reaching for the de-esser to undo your own boost. Some broadcast microphones have a switch that does exactly this lift, which is a reasonable hint about how useful it is and how fixed the amount usually is.
5. A compressor
The gap between your quietest and loudest words is your dynamic range, and a compressor makes that gap smaller. It turns your loudest words down, then raises the whole signal to make up the difference. You set your level once and your quiet words stop getting lost while your loud ones stop jumping out.
It also moves you forward. The mix is everything a listener hears at once: you, your game, your music. A voice with a narrow dynamic range sits in front of that rather than disappearing behind it whenever you relax. How far to take it is in compression ratio explained.
6. A limiter
A limiter is a ceiling. Your voice cannot get louder than the level you set, no matter what you do in front of it. It exists for the laugh, the shout, the sneeze: the one moment that would otherwise arrive far above everything you set up for and clip.
A ceiling only means something when nothing can raise the signal after it, which is why it goes last. Put the compressor after the limiter and the compressor's makeup gain lifts the whole signal back over the ceiling you just set.
Building it in Patchd
Add the six nodes to your mic channel in that order and you have the chain. Nodes are dragged into the order you want, so if you add one in the wrong place you move it rather than starting over. All six are free: High Pass, Expander, De-Esser, EQ, Compressor and Limiter are part of the practical DSP set on every channel, on both tiers.
Starting points rather than settings to copy. High-pass around 100 Hz. Expander threshold just under your quietest speaking level, so it only acts in the gaps. De-esser aimed at the band your S sits in, with the gain reduction meter moving only on an S. EQ lifting 3.5 kHz by a few decibels. Compressor set so the meter shows a few decibels of reduction on normal words rather than pinning. Limiter ceiling a decibel or so below the top.
Then stop. The reason this chain works is that each effect has one job and meets a signal the one before it already cleaned up, which is the general rule in effect chain order.
The takeaway
High-pass to drop the rumble, expander to quiet the gaps, de-esser to catch the S, EQ to lift the range that carries the words, compressor to even the level and bring you forward, limiter to stop the one loud moment. Six effects, each reacting to what the last one produced, which is why the order matters as much as the settings.
Patchd is a Windows audio mixer with this whole chain free on every channel. It is in open beta now, and if you clear the terms your invitation arrives straight away.