Whether you’re a music producer or a recording engineer, knowing how to remove background noise in REAPER will help you save time and, often, money!
While we must avoid background noises at all costs, some recordings may not have the luxury of a controlled environment. For example, you might want to digitize one of your old cassette tapes or use a recording from a field trip. Such recordings will have some noise that just isn’t avoidable. However, specialized plugins and tools can help you eliminate such noise easily.
If your regular mic and instrument recordings are noisy, you might want to check out my tips for noise prevention. Even the best noise reduction software creates artifacts when processing. So, see if you can avoid recording noise at all.
How To Remove Background Noise in REAPER?
REAPER offers you two ways to remove background noise. You can use the built-in plugin ReaFir’s Subtract mode to eradicate a recorded noise profile, which is the easiest and most automatic method. Similarly, you can use the DAW’s spectral edit mode to remove hums, buzzes, clicks, breaths, etc., manually.
Another option is a third-party plugin, which you can apply per media item or track in REAPER. While this method isn’t a native option for REAPER, it can be crucial for certain projects. So, we’ll explore the best practices when using third-party tools in this article.
Now that we understand the basics of how to remove background noise in REAPER, let’s get into the methods in more detail:
Remove Noise Using ReaFir
ReaFir is an FFT-based dynamics processor. Fast Fourier Transform (FFT) is an audio measurement and analysis method that divides an audio signal into individual spectral content. So, ReaFir can divide the audio spectrum (frequency data) into small segments for individual processing. The FFT Size parameter controls how coarsely or finely the plugin divides the audio spectrum; you could think of it as the bands in a graphic EQ.
ReaFIR has four modes: EQ, compressor, gate, and subtract. The first three are self-explanatory and let you process the different parts of the audio spectrum. Furthermore, FFT processing means ReaFir’s EQ is linear-phase, which you might also be interested in since REAPER’s ReaEQ isn’t linear-phase.
Today, we’re interested in the Subtract mode since it’s the automatic noise removal feature. You’ll need a sample of the noise (or any other continuous sound in the audio) without any content you want untouched. So, it’s always a good idea to capture the sound of your environment before or after recording your material (vocals, instruments, sound effects, etc.).
If you do not have a clean sample of the noise, try to find an isolated part in your recording with even a second of the noise in-between takes. Once you have found a noise sample in your recording, follow the steps below:
- Select the audio item you want to process. This item must contain the noise sample. If your noise sample is in a different item, select that item instead and follow the steps.
- Press Shift + E on your keyboard. Or right-click the audio item, find Take, and click “Show FX chain for active take.” We’re using the take FX chain instead of the track FX to avoid creating artifacts in the items that do not require processing.
- The FX selector window will open if your audio item doesn’t already have an effect plugin. Find Cockos in the left pane and double-click ReaFir on the right pane. If you already have an effect in your audio item, double-click on the empty area on the left pane of the FX chain window and add ReaFir as described in this step.
- Change the Mode to Subtract. Set the FFT Size depending on your CPU – a higher FFT Size consumes more CPU. However, I recommend using at least 4096, if not higher. Note that a higher FFT Size might require you to play your noise sample longer for ReaFir to generate a workable profile.
- Position your edit cursor so that it plays right where the noise starts. If you need to listen to the audio to determine the start, bypass ReaFir by turning off the checkmark at the top right of the effect window while you play the audio.
- Turn on “Automatically build noise profile (enable during noise)” in ReaFir. This option instructs ReaFir to record a noise profile.
- Hit play or the spacebar, and make sure you stop it before your actual recording starts. I like to place a time selection over the noise with looping turned on in the transport bar to ensure I don’t accidentally play the vocal or instrument recording. Even with the loop turned on, try not to play the audio more than once, as repetition compounds the profile amplitude.
- Once you’re done, uncheck the “Automatically build noise profile” option and leave it. If you make a mistake while building the noise profile (repeating the playback or playing over the wrong segment), uncheck the “Automatically build noise profile” option and re-enable it. Doing so resets the noise profile recording.
Your audio item should have no noise when you play it back. However, you might notice some artifacts if the noise is significant. Here are some tricks to help avoid artifacts:
- Hold control or command and drag holding the left mouse button up or down on the graphic display to change the noise profile level. It acts like a mix control. Alternatively, the mix knob at the top right of the effect’s user interface can be used. Lowering the noise profile level brings in some noise to keep the sound natural. Sometimes, I even edit the graph manually to avoid extreme lows or highs in the frequency spectrum.
- Right-click the graph, find “Quality settings,” and click “Reduce artifacts.” Sometimes, this option can be useful to retain some of the natural sounds in your recording. As always, trust your ears to decide whether it’s useful for your recording!
- So far, we’ve talked about a single audio item. If multiple items share the same noise, copy the ReaFir instance to each item by dragging and dropping it from your processed item’s FX chain. Alternatively, you can drag and drop the FX icon above the audio item, although doing so will copy the entire FX chain (a non-issue if you don’t have any other take FX). Make sure you have “Per-take FX” enabled for “Media item buttons” in Preferences > Appearance – Media.
I prefer this workflow because you can change the noise profile and customize ReaFir’s settings on each audio item if necessary. However, it can be tedious if you have sliced one audio item into many. In this case, please add ReaFir as a track FX instead.
Note that when you use ReaFir as a track FX, you can get a noise profile from any audio item in the track. Likewise, it will affect every audio item. If you need to separate two segments of recordings, you can use two tracks grouped by a folder track, which will contain the mixing effects like reverb and compressor. Here’s an example:
Remove Noise Using Spectral Editing
Spectral editing is a feature in REAPER that uses FFT to divide an audio item’s frequency spectrum into thousands of bands, allowing you to process the dynamics of each frequency band individually. It’s like ReaFir at its core. However, it’s a manual process and much more visual.
Since spectral editing is manual, removing complex noises like equipment hiss or long wind noise is impossible. However, it’s an excellent tool for noises that either do not change in frequency rapidly (hums, electronic interference, and motorboating effects) or are short bursts (lip smacks, breath noise, and thumps). Remember that ReaFir will likely give you a better result faster if you have a clean noise sample. So, only use spectral editing if there is no distinct noise sample.
Here are the steps to remove noise using spectral editing:
- First, we need to see the spectral view of our audio items inside REAPER. So, open the View menu and click “Peaks Display Settings.”
- Change the “Display mode” to Spectrogram in the window that opens. You’ll notice that your audio items now display the spectral view using FFT. So, you should be able to see the humming or similar noise in yellow or green. Higher-frequency sounds are on the upper part, whereas lower frequencies are on the lower.
- Since the spectrogram is linear, it’s difficult to see the lower frequencies. If your noise is of middle or low frequency, adjust the “freq log” slider until its spectral data becomes clear.
- Place a time selection over the noise by dragging your left mouse click over the time ruler. Press Alt + S or Option + S to toggle Snap for a finer selection. However, the time selection doesn’t have to be perfect.
- Right-click the audio item, go to “Spectral edits,” and select “Add spectral edit to item.” It will create a rectangular box called a spectral edit over your time selection. Alternatively, you can hold Control + Alt or Command + Option and left-mouse-drag over the item to draw a new spectral edit:
- Hold Control + Alt or Command + Option and click on an empty area on the arrangement page to remove the time selection if you created one.
- Now, you can reposition the spectral edit by dragging it around. Or you can drag its top and bottom edges to change the frequencies while keeping its position in time intact. Furthermore, you can hold Control or Command and left-mouse-drag along its top or bottom edges and corners to edit its shape. Holding Alt or Option and dragging along the edges deletes your freehand modifications. You can erase the last node in your freehand drawing to create a straight line from the preceding node, as shown in the gif below:
- Once you’ve isolated the noise with a spectral edit, use the knobs to reduce its volume or change the dynamics. Four knobs allow you to fade in and out the spectral edit’s mix along both frequencies and time. Similarly, the additional knobs on the right side of the spectral edit control the edit’s gain, compressor, and gate. You can adjust the compressor’s threshold and ratio, whereas the gate features threshold and floor controls. Often, the compressor gives much smoother results than reducing the gain, especially when combined with fades.
And that’s all there is! After some practice and experimentation, you can completely remove various noises using spectral editing. However, here are a few tips that should help you get there quicker:
- REAPER limits your spectral edit’s minimum length depending on your FFT size. So, if you want to address an extremely short noise, such as a mouth click, thump, etc., use a small FFT size like 64. Here’s an example with a mouth click:
- The FFT size plays a huge role in spectral editing’s efficiency. So, try different sizes to determine which works best for you. A general rule is to use higher FFT for longer noises with complex frequency content while keeping it smaller for short blips or sine wave-like hums.
- You can apply a spectral edit on a single audio channel or all channels by right-clicking the edit and changing its channel options. Targeting the required channel alone results in fewer unwanted artifacts on the other channel(s).
- In some cases, isolating the noise by removing your actual recording using spectral editing is a better idea. Then, you can use the isolated noise to generate a noise profile in ReaFir. For example, isolating a single electric guitar note with an FFT compressor is much easier than removing the pickup noise.
Gate Noise Using ReaGate
A gate is an audio plugin that turns down the output when the input is below a certain volume level called the threshold. For example, there will be segments in a song where you don’t have to sing. A gate will ensure such segments are perfectly silent by turning down all the noises from breathing, handling, moving around, electronics, etc.
ReaGate is a comprehensive gate plugin available in REAPER. It allows various parameter customization ensuring you have complete control over the audio. Similarly, it can close and open the gate with a side-chain input. Here are some of its main features:
ReaGate’s pre-open parameter allows you to open the gate before the audio starts, which creates clean output but with added latency. Once the audio plays, the attack parameter changes how fast the gate opens, while the hold slider alters how long the gate stays open even though the audio is no longer above the threshold. Similarly, the release parameter changes how quickly the gate closes, allowing you to create a fade-out.
There can be two thresholds in a gate: the opening threshold and the closing. The opening is the threshold to change from silent to playing. Conversely, the gate silences the audio when it’s no longer as loud as the closing threshold.
The hysteresis slider in ReaGate modifies the gate closing threshold. At 0 dB, the opening and closing thresholds are the same as what you’ve set using the vertical slider on the left. However, if you turn down hysteresis to -5 dB, the closing threshold is now 5 dB below your actual threshold. So, if your main threshold were set to -20 dB, the closing threshold would now be -25 dB. Turning down hysteresis is useful when ReaGate keeps silencing your audio even though your performance (vocal release, soft notes on an instrument) isn’t over.
- Send MIDI
The checkmark to send MIDI at the bottom of ReaGate allows you to trigger a user-assigned MIDI note based on the gate’s opening and closing. This feature is useful to trigger drum notes using audio, for example. So, you could use it for drum replacement or even to perform drum parts. I’ve used it to tap on a table with a mic to trigger various drum sounds!
- Input Filter, Noise, and Inverting
ReaGate sports a high-pass and low-pass filter that controls its input. Typically, you wouldn’t want to use this feature, but it’s great when you have to trigger the gate with a certain frequency (like the kick drum) in a mixed track.
Likewise, ReaGate has a noise slider that adds white noise when your audio plays. It might add some live-recorded feel to a MIDI recording. And finally, you can invert the gate, which is great for ducking audio in EDM when used with the side-chain input (kick).
Remove Noise Using Third-Party Plugins
Occasionally, you’ll encounter noises requiring more sophisticated tools than those provided in REAPER. Furthermore, third-party plugins can save you time and effort in bigger projects. So, this section will focus on how to use such plugins in REAPER to achieve desired results. You can find some of the best noise-reduction plugins in this compilation. Also, I recommend checking out Acon Digital’s Restoration Suite, a complete set of audio restoration tools similar to iZotope’s RX 10.
When To Use Take FX
Most noise reduction plugins, such as Klevgrand Brusfri, work like ReaFir: you record a noise profile, and the plugin removes the noise. Third-party plugins may have additional features like machine learning and sophisticated algorithms to provide audibly more natural results than ReaFir’s clinical approach. Still, the core method remains the same. Hence, add your third-party plugin as a take FX if you have multiple recording takes in the same track.
Some noise reduction plugins use ARA to access your audio, although it’s rare. Using take FX for such plugins generally gives better and more predictable results in REAPER. Hence, I recommend the same for other ARA plugins like Melodyne too.
When To Use Track FX
Track FX is much less CPU-intensive than take FX, especially when you require multiple instances of the take FX. So, if you’re processing a single or similarly recorded take sliced into multiple items, you should add the third-party noise reduction plugin as a track FX.
Some noise reduction plugins do not use a noise profile at all. These use special algorithms or machine learning to recognize common elements like mouth-clicking or hissing sounds. Conversely, some have algorithms to recognize and isolate the vocals, which automatically eliminate the noise without a user-assigned noise profile. A good example is Acon Digital’s Extract:Dialogue. You can use such plugins as track FX, although not if you only want to apply the effect to select items.
Pre-Editing Noise Profiles
ReaFIR allows users to easily edit the noise profile by drawing or modifying the curves in its frequency graph. However, most noise reduction plugins utilizing noise profiles do not allow that. This limitation can sometimes fail the processing from reaching its potential.
Hence, I recommend copying your noise sample to an empty section in your arrangement and recording a noise profile after processing it. If the noise is mostly in the high frequencies, try adding an EQ above the noise reduction plugin to cut all the middle and lower frequencies. Making the plugin analyze only the frequency you allow is an excellent way to pre-edit the noise profile and ensure artifact-free processing.
In this article, we’ve discussed how to remove background noise in REAPER. I have described three methods that use REAPER’s native tools: ReaFir, Spectral Editing, and ReaGate. Furthermore, I’ve written some good practices and tips for using third-party noise reduction plugins inside REAPER.
I recommend using ReaFir for most generic post-recording purposes. Similarly, spectral editing is an advanced tool, learning which gives you more and more uses as you discover them. It’s invaluable for editing out mouth clicks, hums, and similar noises requiring specialized audio restoration plugins.
I find ReaGate useful for live performances since I prefer using take-volume envelopes to gate my recordings manually. That being said, ReaGate does save some time, even for recorded audio. And often, automatic gating can be satisfyingly adequate.
And that brings us to the end of this article. While there is no blanket way to remove all sorts of noises in an audio recording, learning about multiple tools can help you understand which method might be the best approach. So, I hope you’ve found this article useful for learning some new skills and tricks. Happy music-making!