Voice Data Partnerships

Make your voices heard

We want to provide lifelike voices for as diverse a set of languages, accents, and use cases as possible.ElevenLabs Voice Data Partnerships make this possible – share your voice data and help shape the future of AI speech technology.

How it works

Step 1
Submit Interest
Tell us about the data you have and how you’re intersted in sharing it.
Step 2
Share Samples
If we're a good fit, we’ll ask for samples to see how we can make the most of your data.
Step 3
Choose Permissions
You decide how your data is used: for training our base models or fine-tuning something for you.

Voice data we want

We're particularly interested in data that reflects the diversity of contexts in which humans speak.


Human-to-human conversations, including casual chats, call center recordings, job interviews, etc.


Audiobooks and all other forms of human reading.

Unique Characters

Yes, those cartoonish ones too! The more creative, the better.

News Reporting

Voices from news studios and on-site reporting.

Interested in becoming a voice data partner?


How high quality does my data need to be?

We don’t have minimum requirements. Voices should be crisp and understandable, and background noise should ideally be no higher than a typical human-to-human conversation. As a reference, anything at or above 44.1kHz sample rate would be considered good quality for us, though for phone calls we can work with as low as 8 kHz.

Can Eleven help me clean and prepare my data?


Do I need to provide transcripts as well?

No. But if you have high-quality transcripts already, please submit them along with the audio.

Will my data be shared with anyone else?

No. We want this to be a true partnership and give you 2 options:
  • We use your data to help our train models to perform better for your domain. This benefits everyone, but nobody has access to your data.
  • We use your data for something specific to you, e.g. fine tuning.
  • Is my dataset too small to be interesting?

    We don’t have minimum requirements. Even small amounts of data can help us better serve low-resource languages and accents.

    I’m not sure about the licenses I have for my data – should I still apply?

    Yes! We can help you figure this out.