Plosives are often a nightmare in recorded vocals. Whether it be in an interview or a recording of a singer. These can be hard to remove, especially if you think the way to go is compression, which is what I initially tried, when figuring out how to remove them. The trick is not compression, it is way easier than that!
For those that don't know, a plosive is the rough popping sound that often comes with the letters "p" and "b". Try putting your hand in front of your mouth and saying "perhaps people prefer Postman Pat", you can feel the air hit your hand as a pressure wave on the "p" sounds. The start of this video sums it up quite nicely.
Overly loud plosives usually sound like loud, often distorted pops. It can be seen below as the big spike in volume in the middle of the recording.
I'm going to use a method, derived from ADSR volume envelope theory. On a synth, if you increase the attack, it doesn't hit as hard. If you lower the attack to zero, the note kicks in harder at full volume.
For a long time, I tried to remove plosives using stupidly complex EQ curves, compression and noise reduction and realised that it was all stupid and didn't work, it also came with a downside of ruining the rest of the audio. So I decided to figure a way to use volume enveloping.
As it is not a note we are playing, we cannot have the envelope kick in at every sound so I decided to use the split function in Reaper combined with the audio fade in feature, this effectively faded in the plosives after the loud pop, keeping the essence of the consonants while eliminating the horrible sound.
The first thing to do is remove all the low frequency noise. The human voice tends not to drop below a few hundred Hz so all the sub-frequencies and much of the bass can be high-passed out. Seen below is the sort of cut I used, just below 200Hz. Remember female voices tend to be higher pitched so more of the low end can be cut out.
After that, we want to start editing out the actual pop. This requires fine work and plenty of zooming in on audio so I recommend disabling the Snap feature in Reaper so that you can move the cursor exactly where you want it.
Next, we want to zoom in on the audio and set the cursor to the very beginning. You can see below that the audio is jagged and inconsistent compared to the note it develops into. Some of this will have been cut out with the EQ, but on playback there is still a definitive distorted plosive.
What we want to do to this plosive is fade it in gently. Reaper's Fade in function occurs at the start of an audio clip and so we will split the audio clip at the cursor, right at the start of the plosive so that it becomes the start of a new clip. (Top Tip: The "S" key splits the selected item at the cursor)
So now we have the rogue plosive as the start of a new audio clip, if you hit play, it will still play seamlessly over the split. We want to adjust the Fade-In of the clip now. If you look above, there is a red line going roughly vertical at the start and end of the clip. This is the volume fade in. By hovering the mouse over the top of the Fade in, the cursor changes and you can drag the fade along the track to adjust the length of the fade in.
What I did was to adjust the fade-in a little bit past the plosive as above. Now, this fade in curve is very steep before levelling off. We want a slower start with a rapid end so the note kicks in immediately after the plosive.
This is easy to change and is demonstrated below. The Fade-In (red) line gets right clicked and a curve menu appears. I like to select the one that looks like an exponential graph as this has yielded best results for me.
Give it a listen and you will hear the distorted pop has gone. It is a good time to fine adjust the Fade-In time now so you get maintain the "p" without the distorted pop appearing again.
Finally, all that is needed to be done is glue that split together by selecting both audio clips and right clicking to select "Glue Items"
Now repeat process for every distorted pop and enjoy the fixed recording.