Cleaning up noisy audio is one of the most common problems in video editing. Trade shows, expos, busy offices, and outdoor shoots with lots of wind noise all introduce background noise that can overwhelm dialogue. DaVinci Resolve Studio includes an AI Voice Isolation feature that can dramatically improve audio recorded in noisy environments. So follow along as this guide explains exactly how to use it, and what common mistakes and problems to avoid.
Takeaways:
- What's the best way to use AI Voice Isolation in Davinci Resolve Studio.
- Full process for using Voice Isolation in Davinci Resolve.
Table of Contents
What Is AI Voice Isolation in DaVinci Resolve Studio?
AI Voice Isolation is an audio tool available only in the Studio version of DaVinci Resolve that uses machine learning to separate human speech from background noise. Unlike typical noise reduction, it focuses specifically on dialogue and removes surrounding sound such as crowd noise, hum, or ambient chatter.
Voice Isolation works exceptionally well when footage was recorded in loud environments and cannot be re-recorded.
DaVinci Resolve Voice Isolation: How to Remove Background Noise from Voice in DaVinci Resolve
Instead of applying voice isolation to individual clips, it is usually better to apply it to an entire track. This is ideal for A roll or voice-over tracks.
- In the timeline, click on the audio track header on the left.
- Open the Inspector panel.
- Make sure you are viewing track-level controls, not clip-level controls.
Enable AI Voice Isolation
- In the Inspector, find AI Voice Isolation.
- Toggle it on.
By default, the strength is set to 100 percent. While this removes a large amount of background noise, it can also make voices sound robotic or artificial and sometimes muffled. It may also cause occasional words to cut in and out, which can be more distracting than consistent background noise. So the most important thing you can do is to fine-tune it manually.
Finding the Right Voice Isolation Level
Instead of leaving the setting at 100 percent, reduce it gradually.
- Lower the percentage bit by bit.
- Listen for a balance between clean dialogue and natural ambience.
A good starting point is around 50 to 55 percent. This keeps some background presence so the environment still feels real, while allowing the voice to remain clear and dominant. The goal is not total silence, but reduced distraction.
Normalising Audio for Consistent Levels
Once the noise is controlled, normalise the audio levels.
- Right-click on the audio clip or track.
- Choose Normalise Audio Levels.
- Set the target to the YouTube standard.
- Leave normalisation as relative if the audio was recorded with the same microphone and settings.
This makes sure consistent loudness is set without changing the character of the sound.
Enhancing Dialogue with AI Dialogue Leveller
- Click on the track again.
- In the Inspector, enable AI Dialogue Leveller.
This feature boosts quieter speech and evens out volume differences across the track.
Optional Background Reduction
Within the Dialogue Leveller options, enable background reduction to further suppress remaining noise. This works well alongside voice isolation, but should be used subtly to avoid artifacts.
Optional Final Audio Polishing
For additional refinement, open the Mixer panel. From here, you can apply dynamics such as gentle compression, adjust the equaliser, or use preset finishers. Many editors find the male lav finisher preset works well for spoken dialogue without much tweaking. The entire process is completely up to you and your end product requirements, so fool around with it until you get it exactly how you want it.
