Tips & Tutorials | October 7, 2015
There are many common audio problems specific to dialogue and voiceover recordings that you can fix quickly and easily in RX 5 Audio Editor, like plosives, mouth clicks, inconsistent levels, and more.
Pronounced plosives on the “P” and “B” sounds can be seen very obviously on the spectrogram by right-clicking on the frequency scale and switching into “Extended Log.” The RX 5 Advanced Audio Editor has a powerful new tool for removing these all at once: the De-plosive module.
Here, we have just three simple parameters to adjust. Sensitivity determines what will trigger the plosive removal, Frequency limit will define the highest frequency that will get removed by the tool, and Strength will decide how extreme the removal should be. This process can be conservatively applied to large selections, or you can select just the problem areas and process them more aggressively without artifacts. Like all the RX processing modules, there are carefully designed presets to help you get started.
You can also address the subtle mouth noises that are present in many vocal tracks and voiceover recordings. The De-click module in RX Audio Editor can automatically detect these clicks and remove them, without affecting the intelligibility of the voice. There’s even a preset to get you started quickly with this particular use case, "Remove mouth clicks."
Inside the De-click module are the parameters that can be adjusted to suit all different kinds of clicks, crackles, and pops. For a use case like this, you would be best served by selecting the “M-Band random clicks” algorithm. There are two parameters to be adjusted to get the best possible sound for this particular voice: Sensitivity and Frequency skew. A Sensitivity setting between 1.5 - 3 should be about right for this case, and we’ll set the Frequency skew around 4 because these clicks are mostly present at higher frequencies, and we don’t want to remove important components of consonant sounds accidentally.
You can hit preview to hear how your settings will affect the sound. If you check “Clicks only,” you can hear just the sounds that are being removed, to ensure you aren’t taking away too much.
Once we have the settings dialed in, we need to make sure we’ve unchecked “Clicks only” and click “Process.”
With plosives and clicks reduced, we’ve removed some of the components that are most likely to cause spurious peaks in our audio track. However, if we look at this audio in the waveform view, you may notice that there is a slight difference between the top and bottom of these waveforms. This is completely natural for the human voice, but it can be fixed to provide extra headroom so that you can achieve more gain before clipping in broadcast scenarios. Open the Channel Ops module and select “Adaptive Phase Rotation,” then click process.
If there is any distracting background noise, we may want to remove that before making any adjustments to the overall level. With the entire audio file selected, bring up the De-noise module. There are many presets to help you get started. For dialogue or voiceover, try “light auto denoising,” which employs the automatic Dialogue De-noise module, and preview.
You can adjust the settings to tailor this to your audio. Threshold controls the point at which De-noise is engaged, and reduction controls how much De-noise is applied. For substantial amounts of background noise, sometimes two passes of de-noise with lower reduction amounts works better than one pass at a large reduction amount. Dialogue De-noise is optimized for voice, and will automatically adjust across the spectrum to tailor transparent noise reduction to the audio.
Now that you’ve removed the most distracting noises from this dialogue, you can level out the audio for a strong, even delivery without having to use excessive compression. Do this by selecting the entire file again, and opening the Leveler module.
Here, RX Audio Editor will analyze the dynamics of the dialogue track and adjust the clip gain to even out large dynamic changes, like an automatic fader ride. Before implementing this, add the clip gain control to our view by clicking on the view menu and selecting clip gain, or by pressing Command+G on a Mac, or Control+G on a PC.
You’ll see that you could make manual gain adjustments here if you wanted to, but in this case let’s use the leveler and then come back and make any adjustments we want after the fact.
Back in the Leveler module window, you can first view current RMS statistics for your selection. You can set a target RMS level for the leveler to work towards. There is a control to adjust the Responsiveness of the leveler; a lower setting will make more aggressive changes to the level, useful if the audio has many sudden variations, and a higher setting will result in smoother behavior, leveling words or phrases rather than individual syllables. There is also a control to Preserve Dynamics, where a lower setting will preserve less of the original dynamics and a higher setting will preserve more of the original dynamics.
While you’re leveling out your dialogue, you can also reduce “S” and breath sounds, while ensuring these don’t get inadvertently boosted in the process of leveling.
Once you have the settings you want, click “Set Clip Gain.”
You can see the results on the clip gain trace, and even go in and adjust these nodes individually to taste, or whole sets of nodes by making a time selection, and holding down the Shift key while dragging a node. You can also remove nodes by selecting the area of audio, right-clicking on that area and clicking, “Remove Clip Gain in Selection.”
These are all steps you can take before mixing to dial in a sound that fits with the rest of the program material. The examples shown here involved a number of different steps, and if you have to process similar audio frequently with the same processing chain, you can build macro-level presets using the Module Chain function.
RX 5 Advanced Audio Editor is the flagship component of the RX Post Production Suite.