How to Create Your Own AI Voice in DaVinci Resolve (Voice Convert)

Davinci Resolve recently got a ton of handy AI features and other upgrades in version 20, and this guide will take you through the steps for creating your own AI voice model in DaVinci Resolve using a high-quality voice recording. Keep in mind, this voice can't be used for Text to Speech, it can only be used to fix sections of bad audio by swapping out the existing audio with an AI version. While you can try swapping out your original audio with someone elses voice it generally sounds like crap...

How to Create Your Own AI Voice in DaVinci Resolve (Voice Convert)

There are two parts to this process, The first part will show you how to clone your voice for future purposes. The second part will show you how to use your newly created voice to enhance and fix sections of bad audio. Both are fairly quick and easy, but as with most things in Davinci Resolve, this will require decent hardware and some patience. This feature is also only available in Davinci Resolve Studio not the free version.

PART 1: Cloning your Voice in Davinci Resolve.

• Start by selecting a high-quality voice clip. Ideally, this should be a clear recording of your voice, such as a podcast or studio session. Place this clip into the Media Pool in DaVinci Resolve.
• Right-click on the voice clip in the Media Pool and navigate to the AI Tools menu. From there, select the option labelled "DaVinci AI Voice Training."

• A pop-up window will appear with information and legal requirements. Read the message carefully. You must have legal rights to use any voice recordings you are submitting for training. After reviewing, click accept.

• DaVinci Resolve will prompt you to download the voice model training module. This is not included by default and must be downloaded separately. The download is free and approximately 4 GB in size.

• Once the training module is installed, you can assign a name to your voice model (for example, "ME"). You will also have the option to adjust training accuracy. After setting your preferences, click "Start."
• Voice model training can take some time, depending on your system. A small processing window will appear as the clip is analysed. During this process, you can continue working within DaVinci Resolve, although performance may be slightly affected. You can check on the progress in the bottom right-hand corner of the Davinci Resolve window.

PART 2: Using your Cloned AI Voice in Davinci Resolve.

Once the processing is complete, your trained voice model will appear in the Voice Convert list. You can now use this model within your projects to apply your AI-generated voice.

• To do this right right-click on the audio file in your timeline that you want to be repaired with your new AI voice clone, then select Voice Convert.
• Now select your Voice Model from the drop-down Voice Model list. And use the drop-down above this to choose if you want to replace the original file or if you want to add it as a New Track.

• Then you just have to wait for the process to complete.

The cool is pretty decent for fixing up bad audio, but I've found it takes some of the emotion out of your voice. I already have a fairly flat voice, and this makes it even worse. Though I'm sure my Australian accent is part of the issue, since most of the options and systems seem to favour US accents.