Blog

3 Steps to Polished Post Production

In the article on creating a voice clone, you might have noticed that I mentioned that clean audio in your voice sample is so important in the quality of the output. It's an equation that should be familiar to anyone in tech and probably also those on the peripherals - good data in, good data out. But what actually is involved in properly producing an audio file? In this post, we’ll walk through a few of the basic techniques and I’ll give samples so you can hear what each step changes along the way.

Now, necessary disclaimer, post production with audio can get complicated. It is an art and a skill. In the interest of making this approachable, we’re going to simplify to the essentials. But I promise, the end result will be high quality. 

First, you’re going to want to make sure you have a good microphone and recording setup. Now I go into microphones in more detail in this post here: Picking your Perfect Microphone - but what is important for you to understand about your recording environment is that...

Silence is Golden

Think of the ceiling fans you might have running, your central air, or maybe your pet water fountain that makes background noises appropriate for meditation but not an audio recording. Anything you can hear may be amplified. Now there is a trick to this, the front of your microphone is more sensitive than the back of your microphone. So, consider where your noise is and place your equipment appropriately. 

There are also several additional elements you can incorporate into your setup such as an over-the microphone pop filter, or my favorite - a closet full of clothes to help sound proof your area. You might be scratching your head and wondering where a closet full of clothes fits into this so allow me to elaborate on that point. You want your space to absorb sound. Think carpet, think curtains, think clothes. These things will catch sound so your recording stays clean. By contrast, if you recorded in an empty room, with hardwood floors, and exposed windows, the sound bound and reverberate through the space, a quality that will impact what is captured by the microphone. So yeah, you don’t have to be fancy, just think of incorporating muffling elements. And this can be as modest as the clothing already in your closet or you can get fancy and install sound proofing panels. 

Personally, when I record, I turn off any ceiling fans, my air conditioning - which though on hot days can be a trial is worth eliminating that “room noise” on a recording, and sometimes I also have to kick my cats out of the room. A good recording, raw and without any edit should sound something like this:

Step 1: Noise Reduction

Okay, you’ve got your microphone, noise elements are under control, and you’ve completed a recording. Now, how do we polish that audio file so it really shines? We’ll we’re going to start with noise reduction, making sure we clean out any obtrusive noises that may be in our recording without affecting the sound of what we want to capture. This goes back to room noise, and making sure the environment for the recording is as quiet as possible. 

Noise reduction is a two-step process so I have a video recording of what this looks like. In your DAW, you’re going to first set a noise gate. DAW, by the way, stands for Digital Audio Workstation and is fancy-speak for the software where you make edits to an audio file. I personally use Audacity so that is what you will see reflected in the screenshots for this post. 

Setting a noise gate will eliminate anything below a set threshold. You don’t want to make this too restrictive because then you can potentially eliminate the sounds you want and nuances in your voice. I have my threshold set at -45dB which works well across most files. I’ve used the same threshold while recording male voices. However, if you have a really deep voice, feel free to play with this as needed.

After we set our noise gate, we’re going to do just a little more detailed clean up. If you look in your audio file, you will see the waveforms of your voice and the flat spaces of silence. Look for bumps in those flat spaces. In the video, you’ll see me highlight one. I’ve also got a screenshot of what this looks like:

These are often breaths and you should be able to see them. Find a breath, highlight it, and mark that sound imprint. So, you’re going to tell the software: “this is the sound profile I want you to look for.” Then, once that profile is marked, select the whole audio and apply a reduction for the sound profile you have created. A sample of what the audio sounds like after this adjustment is below the video. 

Step 2: Normalization

The next step is normalization. This is making sure the recording adheres to a standard threshold. If you listen to the radio, your favorite music app, or any produced audio, the standard level setting is -3. This is what helps make sure you don’t constantly have to adjust your volume dial up and down to accommodate different recordings with different volumes. 

So, we’re in our DAW with the audio recording and we’re going to set our normalization level to -3.0 dB as per the industry standard. You'll hear a volume difference in our audio sample as a result of this step. 

Step 3: Filter Curve

Admittedly, there is some redundancy here with what we’ve already done. But, the additional refinement you get using a filter curve is why this is an important final step. The redundancy is eliminating any noise below a certain level. I’ve attached an image of what my filter curve looks like and you will see a flat line below 60Hz. Basically sound in that range is irrelevant to my recordings. Technically, this should have already eliminated that with the noise gate in step 1. 

There is a flat mid-range to ensure the fundamental frequencies of most voices and instruments in this range remains untouched through the application of this filter curve, preserving their natural quality. And then, a fun scoop around the mid range, at about 1000 Hz is to enhance vocal clarity and cut out any boxy sounds in the vocals. This is admittedly a female vocal curve and the male example is slightly different. This example of a filter curve is intended to cut out any nasal qualities that tend to come up in female voices. It will work for a male voice, but you might want to test slight adjustments. 

Finally, towards the end of the curve, you’ll see a line that jumps up a bit at the end. This is a slight and gradual boost that adds just a bit of air and “sparkle” if you will to the recording. The final recording you will hear as a result of this step has beautiful clarity and polish to the sound. 

Is that it?

Yes. 

You might be shaking your head and thinking, really only three steps? And though audio engineering techniques can be complex, the simple truth will always be that changes to an audio file should be kept to a minimum. You want to keep the voice as natural as possible and making a lot of broad post production changes is what results in an auto-tuned or robotic quality. 

My word of advice in audio editing is that less is more. You want to set yourself up to get the cleanest original recording possible and then make small improvements on that. This approach will give you the best quality and most natural sound in your vocal recordings.