AuthoringAI: Custom Voices

In AuthoringAI, you can create and use a custom voice. Creating a custom voice matches your tone and pacing with AI training data that creates a facsimile of your voice. While this works for most voices, extremely unusual voices or accents may not come through.


Sections within the Article:



Creating Custom Voices

You can create custom voices within the Custom Voices tab of the Media Library. The Media Library can be accessed from two places: either within a presentation using the Media Library button at the top of the editor, or within the My Content tab of the Content Portal.


If your company has enabled custom voices, and you’re in the author role, you can create a custom voice. Each company has a limited amount of voice slots though, so be sure to check with your company’s admin to ensure you can use one of them.


To create a custom voice within the Media Library, you just need to provide a little bit of information and upload at least one voice sample. 

  1. You begin by clicking the Create button.
  2. Then name your voice and provide a description. It’s especially helpful to provide a description if you plan to share your voice with others.
  3. Under the sharing area, you can see that you can limit the voice so that only you can use it, making it a private voice, or you can allow all authors within the company to use the voice making it a shared voice.
  4. Below that you can upload one or more voice samples. Finally, you must confirm that you have the rights and consents necessary to clone this voice, and confirm you will not use the voice for any illegal, fraudulent, or harmful purpose. You can also review our Terms of Service and Privacy Policy here.

Best Practices

For best results use short samples of a single voice, of at least one minute but not much more than three minutes total. Longer samples do not improve the results and can even detract. 


The AI will try to mimic everything that is in the sample you provide, such as tone, pacing, inflection, breathing, and even any background noise. So make sure there are no unexpected artifacts in the recording and that the style is what you intend. If the sample is monotone and emotionless, the custom voice will be too.


Provide a good sample, and you will get a voice very similar to your own. As mentioned earlier, the AI generates your voice with a combination of patterns it has been trained with and your sample, so if you have an unusual accent or pacing, it may not sound exactly like you.


Let’s take a look at the custom voice narrating this.


Voice Security

Whether you keep your voice private or share it, we’re keeping its control in the hands of its creator. 

  • Author role - As an author you can see, preview, and use the private voices you create, as well as any shared voices. 
  • Company Admin role - As a Company Admin can see and preview all private and shared voices in the Media Library, but just like an author, you can only use private you created along with any shared voices to generate audio. 


When an account is removed, if it was used to create any custom voices, the voices are deleted as well. However any slide audio that was generated from the voice will remain in the presentations where the voice was created, just like any audio that might have been recorded by the individual.

Generating Audio

Your custom voice is available to use in any location where you can generate audio. To access custom voices you just switch to the custom voice library.


Under the voice menu you can see the private and shared voices you have access to. Just like when you create a voice, when you use a voice, you must confirm that you have the rights to use it and that you will not abuse them. 


Custom voices are multilingual. If your slide notes are written in a language we support, you can generate audio in that language. 


That’s right, you, or rather your custom voice, can now speak in over twenty languages!


Was this article helpful?
0 out of 0 found this helpful