Eleven Labs - Capi-Metaverse/Template GitHub Wiki

Eleven Labs

Table of Contents

What is an Eleven Labs?

Eleven Labs is a company whose AI voice generator allows users to convert text to speech with a large selection of voices and languages.

Eleven Labs

Each month they give 10,000 tokens (~10 min audio) to do free conversions. In case you reach the limit, you’ll need to wait till next month or pay one of its plans.

This software also allows voice cloning to create the most realistic digital replica of your voice or access to the Dubbing Studio for more control over translation & timing. In order to access these features, it is requiered to pay a subscription.

How to use it?

Steps

1. Create Account (Optional)

This step is required only if you need access to the 10,000 free tokens. Visiting the website without an account provides a reduced number of tokens.

2. Select properties

In order to generate the desired audio, you can select the language and the voice.

In the first case, you can select from a wide range of languages, including Spanish, English, French and Chinese.

In the second case, you can choose between male and female voices.

ElevenLabsProperties1

If you are logged in, additional properties such as the model used and voice stability are accessible.

ElevenLabsProperties2

3. Generate Speech

Once you have configured the properties, you can insert the text and click the "Generate Speech" button. It will take a few seconds to generate the audio. Once the audio has been created, you can listen to it and, if it’s to your liking, you can download it.

Uses in projects

We have used the Eleven Labs AI to generate the voice of Adam, an NPC who guides users through the Tutorial and DNIRenewal Scenes.

Adam Voice

In these scenes we can hear Adam talk in three different languages: Spanish, French and English.

To achieve higher quality output across these languages, we chose the Eleven Multilingual V1 model. Although it supports fewer languages than V2, it produces cleaner results.

We have selected the following voices for each language:

  • Spanish: Charlie

  • French: Charlie

  • English: Daniel

Finally, the remaining properties were set to the following values:

  • Stability: 50%

  • Similarity: 75%

  • Style: 0%

⚠️ **GitHub.com Fallback** ⚠️