Cloning Custom Voice

[CheekiBreekiJesus]

How to Clone Your Voice with Auto-GPT and Eleven Labs

In this post, I'll share how I managed to clone my voice using Auto-GPT and Eleven Labs. By following these steps, you can create your own fully cloned auto-GPT custom responses.

Prerequisites

• downgrading playsound to 1.2.2 (fixed --speak mode not working for me)

pip uninstall playsound
pip install playsound==1.2.2

• Access to the voice cloning feature of Eleven Labs (starter has 80% off first time)

Steps

setup your Eleven Labs Api key in the .env file:

ELEVENLABS_API_KEY=[_MY_ELEVEN_LABS_API_KEY_]

• Upload your voice samples: Upload up to 25 files (up to 10MB each) to train your clone.
• Create your clone: Once created, prompt it at least once.
• Find your voice ID: Go to https://api.elevenlabs.io/v1/voices and find a list of all available voices to you.
• Use Ctrl + F to search for the name of the generated voice.
• Copy the "voice_id" associated with your generated voice.
• Update the speak.py script: In the scripts/speak.py file, update the voices variable with your voice ID.
• Example: voices = ["_MY_VOICE_ID_"]
• Run the script: Run the scripts/speak.py script and test out your fully cloned Auto-GPT custom responses.

With these simple steps, you can clone your voice using Auto-GPT and Eleven Labs. This can be a fun way to generate custom responses that sound like you. Have fun experimenting with different prompts and customizations!

It's important to note that I do not endorse or condone the cloning of copyrighted voices. The purpose of this post is to explore the technology behind voice cloning in a safe and legal manner. Please use these instructions for educational or personal purposes only, and ensure that you are not violating any laws or ethical guidelines. This post was generated with the assistance of GPT-4.

[BAIS1C]

HI there, I do not see any customised voices in the link v1/voices I only see the standard voices.

[BAIS1C]

OK found solution, the link you porvided does not show the personally created voices, however, if through the documentation, https://api.elevenlabs.io/docs#/ You input your API Key in the try it, section you can pull your own custom voices.

[emregurkn]

You are the man

[sirplus51]

I have custom voiced on elevenlabs only it took me a while to workout the format of the id needed in the env file
eg
ELEVENLABS_VOICE_1_ID=your-voice-id-1

in eleven labs
I have a voice
Albo which was cloned

ELEVENLABS_VOICE_1_ID=Albo <-- Doesn't Work!
ELEVENLABS_VOICE_1_ID=albo <-- Doesn't Work!
ELEVENLABS_VOICE_1_ID=cloned/Albo <-- Doesn't Work!
ELEVENLABS_VOICE_1_ID=cloned/albo <-- Doesn't Work!
ELEVENLABS_VOICE_1_ID=cloned albo <-- Doesn't Work!
ELEVENLABS_VOICE_1_ID=cloned Albo <-- Doesn't Work!

My api appeared to work okay
The result is voice is not found and the voice used is always the default

Below is how to work out the id they are after

Goto
https://api.elevenlabs.io/docs#/
Third option down is
Text To Speech
voices
Access to voices created either by you or us.
Click that and chuck in your API
Click Execute
Look in the generated code block (3rd block down)
Scroll down in the code a fair way
alll the default ones are above the custom ones (at bottom)

What you are looking for is like odLoYy7xsQFzhb3S28Db

[reubenfinley]

@sharpycat , yeah, i'm not seeing the speak.py file either. let me know if you figure it out.

[jacksparrow124]

For the voice, we could also use Whisper API for that right?

[HARRYFOXAUS]

This took me so long to find until stumbling across this and fixed it in seconds. Thank you sir, my little robot is now talking to me in the right voice!

[Soberboi]

How did you update the voice. I tried the voice ID for a custom and premade and neither work. Just getting a default voice.

[HARRYFOXAUS]

yo did you follow this part of the instructions:

"Below is how to work out the id they are after

Goto
https://api.elevenlabs.io/docs#/
Third option down is
Text To Speech
voices
Access to voices created either by you or us.
Click that and chuck in your API
Click Execute
Look in the generated code block (3rd block down)
Scroll down in the code a fair way
alll the default ones are above the custom ones (at bottom)

What you are looking for is like odLoYy7xsQFzhb3S28Db"

Bc that last line of numbers and letters is what the id should look like. it doesn't know the names you give your models, it uses these alphanumerical codes to access the voice data on ElevenLabs