Compression
Your voice rises and falls without you noticing: a laugh shoots up and a sentence at the end of a breath almost disappears. The compressor evens out those differences so the listener doesn't have to touch the volume. Here you'll learn what each control does and some starting settings that work.
What a compressor does
A compressor is an automatic, very fast volume control. When the signal goes above a level you choose, it turns it down; when it's below, it leaves it alone. The result is a voice with less difference between the loud and soft parts: less dynamic range.
Why does it matter in a podcast? Because most people listen to you on the subway, in the car or doing the dishes. If the soft parts get lost in the noise and the loud parts are jarring, the listener ends up turning the volume up and down… or turning you off. A sensibly compressed voice is always understandable, at any volume.
The “close, present” voice of radio is, to a large extent, compression. Stations process the signal heavily so it sounds good on any device. A podcast doesn't need to go that far, but the idea is the same.
The five controls
Almost every compressor, from the one in Audacity to those in any professional software, has the same controls with similar names:
| Control | What it does | Starting point for voice |
|---|---|---|
| Threshold | The level at which the compressor starts to act | So it acts on the loud syllables, not the whole sentence |
| Ratio | How much it turns down whatever goes over the threshold. At 4:1, if 4 dB go over, 1 dB comes out | 3:1 to 4:1 |
| Attack | How fast it reacts when the signal crosses the threshold | 5–20 ms |
| Release | How fast it stops acting when the signal drops back down | 50–200 ms |
| Make-up gain | Raises the whole result to recover the lost volume | Roughly the amount it has reduced |
Some compressors add a knee: a soft knee makes the compression kick in gradually and sounds more natural on voice.
A compressor is not a volume knob
A very common mistake is thinking the compressor “turns up the volume”. It actually does the opposite: it turns down the loud parts. The feeling of a more powerful voice comes afterwards, when you raise the make-up gain on a signal that's already been evened out. That's why, without that last step, a compressed voice sounds quieter than the original.
The gain reduction meter
The most important control isn't a control: it's the gain reduction meter, which shows how many decibels the compressor is taking off at each moment. It's your guide to not overdoing it.
- Between 3 and 6 dB of reduction on the voice's peaks: natural compression, barely noticeable, but the voice comes out more even.
- More than 8–10 dB constantly: it starts to be audible. The voice loses life, and the breaths and background noise rise.
- 0 dB all the time: the threshold is too high and the compressor is doing nothing.
Try it in the simulator above: lower the threshold gradually while watching the meter, and compare the waveform before and after.
How to set it step by step
- Pick a representative section, with soft parts and a laugh or some emphasis.
- Set the ratio to
3:1, the attack to about10 msand the release to about100 ms. - Lower the threshold until the meter shows 3 to 6 dB of reduction on the loud syllables.
- Raise the make-up gain until the perceived volume is similar to the original.
- Compare with the effect on and off at the same volume. If you don't, you'll always prefer whatever sounds louder.
The usual order is to clean up first (noise reduction), then EQ and then compress. If you compress a very noisy voice, the compressor raises the noise in the pauses along with the make-up gain.
Don't squash the voice
The most common mistake when you discover the compressor is overusing it. An over-compressed voice sounds flat, gets tiring after ten minutes and exposes everything that was hidden before: breaths, mouth clicks, room noise.
Don't
Ratio 10:1, a very low threshold and 15 dB of constant reduction: the voice sounds loud, but with no natural breathing and the background noise rising in every pause.
Do
Ratio 3:1, 3–6 dB of reduction on the peaks and just enough make-up gain: the voice is always understandable and still sounds like a person.
With an attack of 1 ms or less, the compressor bites the start of every syllable and the voice loses punch. If the consonants sound dull, lengthen the attack a little.
The final limiter
A limiter is an extreme compressor, with a very high ratio, that stops the signal from going above a fixed ceiling. It's not for shaping the voice, but a safety net: it goes at the end of the chain, after everything else, so no peak goes over the maximum.
For podcasts, a ceiling of −1 dBTP (true peak) is the usual reference. If the limiter is often working more than 2 or 3 dB, the problem is earlier: check the compression or the overall level. You'll see it in detail in loudness and LUFS.
| Tool | What for | Where it goes |
|---|---|---|
| Compressor | Evening out the voice's dynamics | After cleanup and EQ, on each voice track |
| Limiter | Making sure nothing goes over the ceiling | At the very end, on the master mix |
- The compressor turns down whatever goes over the threshold so the voice is more even.
- Starting point: ratio
3:1–4:1, attack5–20 ms, release50–200 ms. - Let the meter guide you: 3–6 dB of reduction on the peaks.
- Always compare at the same volume and don't squash the voice.
- Limiter at the end, with the ceiling at
−1 dBTP.
Your turn: three compressions, one sentence
Apply gentle, medium and exaggerated compression to the same recording, and listen at the same volume to train your ear.
- Choose 30 seconds of your voice with a laugh or some emphasis.
- Export it three times: 2 dB, 5 dB and 12 dB of reduction on the peaks.
- Listen on headphones, adjusting the volume so they sound equally loud, and note which sounds most natural.
Frequently asked questions
What compression ratio should I use for a podcast voice?
A good starting point is between 3:1 and 4:1, with the threshold set so it reduces 3 to 6 dB on the peaks. From there, adjust by ear.
What's the difference between a compressor and a limiter?
A compressor gently reduces whatever goes over the threshold to even out the voice; a limiter is a hard ceiling that stops the signal from going above a maximum level. They're used together: compressor on the voice and limiter at the end.
Why does my voice sound worse after compression?
Most likely you're compressing too much or the attack is very fast. Lower the ratio or raise the threshold until it takes off only 3–6 dB, and compare at the same volume.