Audacity has a Noise Reduction effect which has its default settings. These default settings work well with most audio files but are not ideal for some recordings. This is because the level and type of noise will always vary from one file to another.
You can pick a squeaky door hinge noise when recording today and the next time pick a fan whirring noise. These two are not treated using the same noise reduction settings. Therefore you have to adjust these default settings on some occasions to have the best sounding audio after the noise is reduced.
These settings that are adjusted in the noise reduction process in Audacity are Noise Reduction, Sensitivity, and Frequency Smoothing Bands.
Before you get to this dialogue box with these three settings you need to select a sample of the noise in your audio recording. These settings are only available once you are done with making a selection of the noise sample first. The larger the sample selection, the more effective this noise reduction process will be. Without selecting a part of your file you will always get this dialogue box
The noise reduction process in Audacity has two steps. Step one is getting the noise sample.
Here is the step 1 process:
- Launch Audacity. Import the file that you want to reduce or remove background noise.
- This image has four highlighted areas. These are the breaks or gaps in between speech. In these sections, the ambient noise is much louder and more audible than in the sections with speech.
To get to the noise reduction settings you have to get the Noise Profile first. A noise profile is the characteristic of the noise in your audio. To get the noise profile select any part of your file that does not have speech. The sizeable selection here is either the first or the third section in the image attached above.
- Select by left clicking and dragging the mouse over the section. After this click on the Effects option. Under the dropdown list that opens click on the Noise Reduction option.
This dialogue box will open up.
- Click on Get Noise Profile.
This action will result in the dialogue box closing. Select all of the audio file by double clicking on the waveform or using the keyboard shortcut Ctrl + A (Windows) or Command+A (macOS).
Once again click on the Effects option followed by the Noise Reduction option.
This dialogue that has the noise reduction settings will open again box will now open again.
After step 1 we transition to step two which is readjusting the settings.
Best Settings for Noise Reduction
The highlighted part in the image below has three sliders. These are readjusted to get the best noise reduction effect on your file.
Here are the readjustments that you need to make on the sliders to get the best settings for noise reduction:
Noise Reduction (dB)
This slider is used to change how much volume reduction is applied to the noise level. If your noise level is at -48dB and you drag the slider to a 40dB figure then the noise level will be reduced by 40dB. To get an estimate of this value play a region that has noise and check the level on the playback level bar.
For this first slider, dragging it to the right gives a higher value. Dragging it to the left gives you a lower value. The higher the value the more noise you reduce. Dragging it to the extreme right will remove all of the noise but will sometimes affect the remainder of the file. You will have this underwater effect on the speech. For louder background noises or high-pitched tones, this slider can be dragged past the 40dB mark. For milder noises keep the slider within the 20dB to 40dB range. Lower noises like a constant low hiss can be treated with a figure below the 20dB range.
This second slider is used to gauge which how much of the audio will be regarded as noise. The higher the value you input in this section the more noise you will end up removing. A higher value is obtained by dragging the slider to the farthest right. Lower values may introduce artifacts. Artifacts are unwanted sounds or noises such as crackling or hissing. Aim for a 5 to 10 range.
Frequency Smoothing Bands
This controls how much the noise reduction spreads to other adjacent frequencies. It helps smooth out possible distortion that might occur due to increasing the noise range. For instance, you can be recording and the recording picks up a high-pitched noise from equipment like the computer within the studio. High pitched noise can also be a backup beeper from a reversing car. For such a case that has a high pitched frequency set this slider with a lower value.
The ideal range is 0-6.
Audio properties and quality vary from one file to another. Therefore, whenever you are dragging the slider to the range suggested above always take time to listen back to how your final audio will sound after you apply the changes.
This dialogue box has a preview button.
Following the highlighted parts in the image above you can either listen back to the file with the Noise selection on Reduce or Residue. When you are previewing the recording on Reduce you are listening to audio with the background noise reduced or removed. In this case, you are listening to the voice. On the other hand, the Residue option allows you to listen to the noise that you are removing. This option is useful in determining whether parts of the person’s speech are being cut out as noise.
Before applying the changes by clicking the OK button always ensure this preview setting is on Reduce.