Use an AI Voice Generator to Fix a Mistake in Your Video

March 23, 2023

Chris Lavigne


Did you know you can use a Voice AI to fix a line reading mistake in your voiceover without needing to completely reshoot your video?

It’s true! It’s yet another new Generative AI tool that is changing the game for video producers.

In this post, we’ll show you how to use a new AI tool from ElevenLabs. We’ll also discuss the significance of this groundbreaking technology for video production and how you can start using it.

When a dub can’t fix a flub

If you’re a video producer, has this ever happened to you?

Let’s say you flubbed a line reading or you have to re-record a script change after you’ve wrapped production.

The bad news is that you have no camera or audio gear with you and you’re far away from the location where you originally recorded.

If you try to re-record your voice with a different microphone setup and in a different sound environment, it’s not going to match the original recording.

But thanks to advances in AI, you can create voice models based on different microphone and audio capture situations to match your re-recording with the original recording.

How AI changes the VO game

What’s unique about this technology is that you can use samples or source material to train the AI model to create a voice print that will match your original audio setup, and it’s instantly ready to use.

This means that no matter what your audio circumstance was while you were filming your video — whether you were recording with a shotgun microphone in a studio, or with a built-in laptop microphone during a webinar, or using a USB microphone for a podcast — you can very easily make a custom voice with a sample from whatever audio environment the original recording took place in.

How to use the Voice AI

To do this, head into ElevenLabs to use their “Prime Voice AI” tool.

Over in Speech Synthesis, click “add voice.”

Now, you’ll need to train the model with the voice from your video. So you should export 1–2 minutes of whoever is speaking without any background music and upload that to ElevenLabs.

Once your voice is trained, you can use text-to-speech to type in the line that needs to be updated in the video.

You can even tweak some settings to better match the tone and inflections of the original line reading you need to replace!

When you’re satisfied with the sound, hit generate, download the audio, drop it into your video editing software, and boom! You’ve just saved yourself a reshoot.

Try it out for yourself

We continue to be amazed by innovations in AI technology, and we’re happy to share our findings with you! Give this AI hack a shot in the next video that needs an updated line reading. And be sure to subscribe to our blog to stay updated on the latest video production news like this!

March 23, 2023

Chris Lavigne


Mailing list sign-up form

Sign up to get Wistia’s best and freshest content.

More of a social being? We’re also on Instagram and Twitter.