By Justin Moon

We can't always record audio in the best of environments. Sometimes there is noise from the air conditioning, a vending machine, traffic outside the window, etc. For you audio and video DIYer's and staff members tasked with sweetening less than ideal audio, here is a quick guide to noise and how to reduce it using Adobe Audition.

First, let's identify the types of noise that can be reduced. Noise that occurs in a predictable frequency range, and at a fairly stable volume such as fans, hum, hiss, room noise, etc. can be reduced fairly easily. Noise such as others talking, sirens, or doppler noise that changes pitch as it goes along in time are more difficult and should be noticed during recording if possible, and a retake of that section performed.

Notice also that we aim to "reduce" noise, not "eliminate" it. Think of yourself in a restaurant with a friend. Their conversation is near your ears, and easily perceived, versus the people talking quietly two tables over. That noise doesn't impact your ability to hear your friend. Now imagine the people two tables over being 3 inches from your ear and speaking at a normal volume. Things become a little more distracting. Our aim in noise reduction is to simply push the noise further away, so that the intended sounds are perceived as being closer and the noise is perceived as being further away. Attempts to eliminate noise result in the creation of digital artifacts in the "good" audio that actually becomes more distracting than leaving the noise alone in most cases.

The process goes as follows: Define, isolate, reduce.

Define Noise: Open Adobe Audition, and open your audio file in the waveform editor.

""

In this example we can see there is some fan noise. Luckily, it was recorded without speech over it for a little while before the speech began. In this case it is easy to isolate, but we will look at finding it within the speech itself as well.

Zoom in close on the noise, using the mouse wheel, and select a very small sample of noise. A fraction of a second is plenty. Make sure that the sample contains only the noise you want to remove, and there is no speech or breaths occurring in the sample.

""

Next we will select the “effects” menu, then “Noise Reduction/Restoration” then “Capture Noise Print.”

""

Nothing will visually occur confirming that you have a noise print. Next, go back to the same menu, but this time select “Noise Reduction (Process)…”

""

Once this is selected, the noise reduction effect will open.

""

I find that shorter samples work best in noise reduction. If you received a message asking for a longer sample when selecting noise, you can click the advanced dropdown, and change the FFT size to a lower number, and repeat the noise print selection with a smaller size.

Click the “Select Entire File” button.

""

This will allow you to preview the noise reduction as you listen. You can press the space bar to play the audio and make adjustments to your liking. As you move the two sliders “Noise reduction” and “Reduce By” you will notice what the effect is doing. Notice the harsh and digital effect that moving the “Noise Reduction” slider all the way to the right has. This is noise elimination, we are just trying to reduce it. Remember, just push the noisy people further from your ear in the restaurant.

For my example, 67% and 19.6db on the sliders were plenty to reduce this noise almost entirely.

""

Once you are happy with the results, click “Apply.” If you are satisfied, you can then save the file.

""

We can see that compared to the first image, the noise has been reduced significantly, while leaving the speech content largely unaffected.

But what if there isn’t a clear picture of the noise in the beginning of the file like this example? That’s mainly why I like a small sample size in the effect setting. In this case, we need to zoom into the waveform and find a pause in the speech where the noise is occurring on its own. And this can sometimes be only a fraction of a second.

""

Here I have selected a pause in between words that contains only the noise. You can repeat the process of capturing the noise print from here.

The process for reducing noise is the same for all noise you intend to reduce. Some noise is very easy to eliminate, such as hum from electrical wires, some fan noises, etc. Others can be more challenging. It is best to record in an environment free of such noises, but we know that isn’t always the case. If you have any questions about this or other audio/video processing, email Justin Moon, Manager of Media Production at This email address is being protected from spambots. You need JavaScript enabled to view it..