Skip to main content

Custom Voice Cloning

This Cookbook takes about 10 minutes. You'll register an account with ElevenLabs and use it to create a custom voice for reading text. This is extremely useful for accessibility, if you've lost your voice and have recordings of it.

Why Custom Voice Cloning?​

Custom voice cloning is a powerful tool with sharp edges. It can be used to create deepfakes, impersonate people, and create fake news.

It can also be used to create accessibility tools.

Telosnex was not planning on adding this feature, but it was requested by a user who lost their voice and wished to use Telosnex to communicate. We decided to add it, but with a strong warning about the ethical implications.

Why ElevenLabs?​

ElevenLabs is a company that provides text to speech services, and the ability to create custom voices.

When you register with ElevenLabs, you can create a custom voice, and then use it for free in Telosnex: all billing will be handled by ElevenLabs. Having your own account also guarantees that your voice will not be used by others. (though, optionally, you can share it with others)

ElevenLabs offer paid monthly plans with steep discounts for a certain number of minutes of audio per month, and then charge per minute after that. As of June 2024, the cheapest plan with custom voices is $5/month and includes 30 minutes of audio. [ElevenLabs Pricing]

1. Register with ElevenLabs​

  1. Go to ElevenLabs.io and register an account.

  2. Once you're signed in, get your API key. As of June 2024, you do that by clicking on the circle in the top right, then clicking on "Profile + API Key".

  3. Set the API key in Telosnex by going to Settings > AI > API Keys > ElevenLabs, and pasting it in the API Key field.

2. Create a Custom Voice​

  1. You can use Telosnex to create a custom voice by uploading an audio file. However, we recommend using ElevenLabs' web interface for this.
  2. Go to ElevenLabs's Voice Lab.
  3. Choose "Instant Voice Cloning"
  4. Follow the instructions to add audio samples. The more you record, the better the voice will be, but you get good results with as few as 1 minute of audio.
  5. At the bottom, confirm you have rights to use the voice.

3. Use the Custom Voice in Telosnex​

  1. Open Telosnex and go to Settings > AI > Text to Speech.
  2. Press 'Use' under ElevenLabs.
  3. Choose your custom voice from the list. The voice's name will be the same as the name you gave it in ElevenLabs.
  4. That's it! Press Test if you'd like. From then on, when you press Read, the custom voice will be used. You can find Read under the latest message in a Chat, or, by pressing the 3 dot menu on any Chat message.
Script your Custom Voice

Make a Script with one step, Dialogue. Set its text to {{prompt}}. Then, hide it. When you use this Script, your prompt will be echoed back in the Chat, with a Read button you can press immediately.