With location dialogue tracks, a steady background noise can make it difficult to bring quiet scenes forward in the mix without also boosting the distracting noise floor. Using the EQ module before the Voice De-noise module in iZotope RX (Dialogue De-noise in previous versions) can help mitigate this constant noise and allow the voice to be heard more clearly and prominently without losing the ambience of the scene. If removing background noise from dialogue is a frequent noise reduction task for you, consider trying the Dialogue Isolate feature in the Advanced version.
Here is an example of some production audio from a boom mic with a significant amount of background noise from a nearby highway.
- Open the audio file in the RX Audio Editor or send it via RX Connect.
- Open the Corrective EQ module [Option+Shift+7].
- Engage a high-pass filter to remove the most apparent rumble and to make any other static filtering gestures before applying the De-noiser. In this example, we also reduced some of the prominent ‘S’ frequencies around 7 kHz and a tonal component of the background noise around 800 Hz.
- Then open the De-noise module [Shift+4]. The De-noise module has two modes: Dialogue and Spectral. We’ll use Dialogue mode for this example.
- Inside the Dialogue tab, we can set the De-noise algorithm to adjust automatically (which is used for sounds that vary throughout the program), or we can manually learn a noise profile that the algorithm can reduce constantly across the program. Since this example has steady background noise throughout, we’ll start in Manual mode.
- Now we’ll Learn a noise profile by selecting a passage of at least one second of pure noise in your audio and clicking Learn.
- The six Threshold Nodes will automatically set themselves based on the noise profile. These nodes represent different parts of the frequency spectrum, and their thresholds can be adjusted (and automated) individually.
- Click Preview and adjust settings to the program material, starting with the Reduction slider and then adjusting multiband threshold nodes if necessary.
- Once you have arrived at the optimal setting for your audio, click Process.
- Insert the iZotope RX Dialogue De-noise plug-in on the dialogue track or bus that requires noise reduction. In some cases you may want to insert an EQ plug-in ahead of the Dialogue De-noise plug-in to filter low frequency rumble and make any other broad EQ adjustments.
- Start in Auto mode. This is designed to dynamically adjust the individual threshold nodes across the frequency spectrum throughout the program to accommodate any changes in the noise profile.
- Play a few seconds of audio for the plug-in to initialize the noise reduction curve.
- Raise the Reduction value until you hear the noise begin to disappear. Try to set the Reduction value to an amount slightly below where it begins to affect the voice itself.
- Raise the Threshold value to set the point at which the noise reduction engages. A higher threshold value may start to suppress low-level components of the voice itself. Strive to find a balance between the Threshold and Reduction values that provides the desired amount of noise reduction with the least amount of artifacts.
- In Manual mode, you can adjust the Threshold individually for each of the 6 multiband nodes to tailor the amount of noise reduction across the Frequency Spectrum.
- You can also Learn a noise profile in Manual mode:
- Find a passage of at least one second of pure noise in your audio.
- Click Learn inside the Dialogue De-noise plug-in (you must be in Manual mode).
- Play the selection back through the plug-in so that it can assess the components of the noise.
- When you stop playback, the threshold nodes will automatically set themselves based on the noise profile.
- Now you can make adjustments to the threshold nodes, master threshold control, or reduction amount to tailor the noise reduction to your dialogue by ear.
If the noise changes throughout the program:
To learn a noise profile that remains constant:
- Use in moderation. Noise reduction artifacts can be more distracting than the noise itself! Be gentle and use multiple passes of De-noise if necessary. This will often lead to a better sonic result than one De-noise treatment with a very high reduction amount.
- Stereo Noise, Mono Signal. If the noise is in stereo, but the original signal is predominantly mono, the Center Channel Extraction feature in the Channel Ops module of RX Advanced can be used to separate the stereo noise from the mono signal.
- Treat the high-mids gently. When in Manual mode, set lower thresholds on the high-mid frequencies, as this tends to be where vocal artifacts are more audible.
- Process Tonal and Noisy signals differently. If the original signal is predominantly tonal, the Spectral De-noise module allows separate control over Tonal and Noisy reduction to reduce artifacts and pinpoint problem areas. Click on the small chain icons underneath the Threshold and Reduction sliders to un-link the tonal and noisy parameters and control them separately.
- Know your noise. Is your noise constant or varied throughout? Is your noise broadband or tonal? RX has customized tools for different types of noise. You can learn more about identifying these different types of noise in the article Identifying Audio Problems.
- Automate each threshold node. All of the parameters of the Dialogue De-noise plug-in can be automated when it is in Manual mode, including the individual threshold nodes, so that you can adjust settings in real-time along with your program.
- Sculpt with EUCON. The Dialogue De-noise Plug-in is EUCON compatible so you can configure any of the parameters to be available right on your control surface for real-time adjustment and automation.