I am about to show you how you can clone any voice in DaVinci Resolve. If you have someone mess up a voice line in your video, you can now clone their voice and fix it so it sounds perfect. This is a total game changer for video editors who need to save a scene where the audio just didn’t hit right.
Before we dive into the technical steps, if you want to learn some of my best editing tips, I have a completely free Edit Videos Like A Pro Guide that you can download right now.
The Real World Scenario
Imagine you are filming a wedding and the groom is reading a letter to his bride. He’s nervous, and he stumbles over a few words. Usually, you’d just have to live with it or try to cut around it. But with the AI tools in DaVinci Resolve, you can actually fix a mistake of virtually any size with very little work.
Step 1: Gathering Your Training Audio
To start, you need some audio of the person whose voice you need to clone. Ideally, you want at least five minutes of them speaking. In my case, I had about eight minutes of the groom speaking because I had a microphone on him well before he started reading his letter. Even if you have less, it is worth trying, but more data usually leads to a better voice model.
Step 2: Training the AI Voice Model
This part is incredibly easy. Right-click on your audio file in the media pool, go to AI Tools, and select DaVinci AI Voice Training. The first time you do this, you will likely need to download some extra files. Once those are ready, click it again, and you’ll see a warning screen. This tool is powerful, and Blackmagic wants to make sure you use it ethically.
Name your voice model (for example, “Paul Groom Voice”) and choose your training accuracy. I recommend choosing “Better” over “Faster” because you want the most accurate results possible. Your computer will then start generating the model. Since this happens entirely on your device and not in the cloud, it might take a while depending on your hardware, but you can keep editing in the background while it works.
Step 3: Recording the New Line
Now for the part that sounds like magic: you need to record yourself speaking the line that the talent messed up. You don’t even need to use the same microphone or be in the same environment. I recorded my “clone” audio in a sound-treated studio with a shotgun mic, even though the original was recorded in a barn in West Texas on a lav mic. It doesn’t matter!
Step 4: Converting the Voice
Import your new recording into DaVinci Resolve and place it on your timeline. Right-click the clip, select Voice Convert, and choose the voice model you created earlier. Hit render, and watch as Resolve analyzes the audio and transforms your voice into the voice of your talent. If you want to take your visuals as seriously as your audio, don’t forget to check out my LUTs sales page for better color grading.
The Moment of Truth
When you play it back, the results are often insane. It sounds like the original person saying the words you just recorded. Now, keep in mind that this won’t automatically match the lip movements of the person on screen. This is where good editing comes in! Simply cover that section with some B-roll, and no one will ever know the audio was manipulated.
This technology is an incredible use of AI because it helps us tell stronger stories and make our videos clearer. If you found this helpful, definitely subscribe for more DaVinci Resolve tutorials and filmmaking tips.