How to Clone Your Voice with Claude

Claude can turn one approved recording into a voice you can reuse for narration, videos, podcasts, and product demos. CreativeClaw handles the private upload and provider connection, while Claude guides the workflow and generates new speech from your prompts.

You can choose ElevenLabs or Cartesia, save the clone on a Character, and call it again by name in a later conversation.

Before you start

You need a connected CreativeClaw account, a clean audio recording, and the right to clone the speaker. Reusable voice cloning is available after a CreativeClaw workspace has made a credit purchase.

Voice cloning requires explicit permission. Do not clone a public figure, customer, coworker, friend, or family member unless that person understands and approves the cloning and its intended use.

Step 1: connect CreativeClaw to Claude

For Claude apps, add the CreativeClaw connector and sign in. For Claude Code, install the CreativeClaw plugin. Follow the appropriate steps in the Claude connection guide.

Step 2: record 1 to 2 minutes of speech

Record yourself speaking alone in your primary language for one to two minutes. Include different emotional tones, but keep each one natural:

• calm and conversational
• friendly and upbeat
• focused and serious
• energetic and excited
• warm and reassuring

Use a quiet room, maintain a consistent distance from the microphone, and avoid background music or another person’s voice. Speak in complete sentences. Your everyday voice, with a little emotional variety, is a better source than exaggerated character acting.

If you need text to read, ask Claude to write a one-minute sample script in your language that moves through those tones. Read it once naturally, without trying to sound perfect.

Step 3: upload the recording privately

Attach the audio file or give Claude access to the local file, then ask:

Use CreativeClaw to upload this audio as a private voice-cloning sample.
Create a reusable voice-only Character named "My Voice" and clone it with
ElevenLabs. This is my own voice, and I explicitly consent to the recording
being processed and cloned for use in my CreativeClaw workspace.

For Cartesia, name Cartesia in the request. If Claude needs an interactive file picker, CreativeClaw can open one. The resulting sample is kept private and represented by an asset ID rather than a public audio URL.

Step 4: choose ElevenLabs or Cartesia

CreativeClaw supports two reusable cloning providers:

Provider	Choose it when you want
ElevenLabs	Natural narration, expressive delivery, and broad multilingual speech
Cartesia	Low-latency Character speech and direct emotion, speed, and volume controls

ElevenLabs is the default when no provider is named, but it is better to choose deliberately. You can also tell Claude about the project and ask for a recommendation before cloning.

Step 5: generate your first speech

After Claude confirms that the Character has a cloned voice, try:

Use my saved CreativeClaw Character voice to generate this speech:
"Hello, and welcome. This message was created in Claude using my reusable voice."
Make the delivery clear, warm, and lightly upbeat.

Claude can revise the script before generating it, and you can request a different pace or emotion afterward. Reuse the Character ID or saved name. Do not create a fresh clone for every script.

Useful follow-up prompts

List my CreativeClaw Characters and tell me which one has my cloned voice.

Use my cloned Character voice for this 30-second tutorial. Speak slowly,
sound confident, and pause briefly between the three steps: [paste script]

Rewrite this paragraph so it sounds natural when spoken, then generate it
with my cloned voice: [paste paragraph]

Keep the result trustworthy

Tell listeners when synthetic speech could reasonably be mistaken for a real recording in a sensitive context. Never use a clone to impersonate someone, fabricate approval, or mislead people about who said something.

CreativeClaw’s voice-cloning terms explain the permission requirements. The privacy policy explains private sample storage and provider processing.

Frequently asked questions

How do I clone my voice with Claude?

Connect CreativeClaw to Claude, provide a clean 1 to 2 minute recording, ask Claude to upload it privately for voice cloning, choose ElevenLabs or Cartesia, explicitly confirm speaker permission, and save the clone on a reusable Character.

Can Claude clone my voice directly?

Claude coordinates the workflow, while CreativeClaw privately imports the sample and creates the reusable clone with ElevenLabs or Cartesia. Connect CreativeClaw to Claude before uploading the recording.

What should I record for a voice clone?

Record 1 to 2 minutes of clean speech in your primary language. Use your normal voice and include a few natural emotional tones, such as calm, cheerful, serious, and enthusiastic.

Does CreativeClaw support both ElevenLabs and Cartesia?

Yes. You can explicitly ask Claude to clone the voice with ElevenLabs or Cartesia. The provider is not changed silently.

Do I need permission to clone a voice?

Yes. Clone only your own voice or a speaker who has given explicit, informed permission for the recording, cloning, processing, and intended use.

How do I use the voice again?

Ask Claude to use the saved CreativeClaw Character by name or list your Characters and select it. You do not need to upload the original recording for every speech.

How much does it cost to clone a voice with Claude?

CreativeClaw currently does not charge credits for the clone creation itself, but reusable voice cloning requires a workspace that has made a credit purchase. Generated speech uses credits based on the selected model.

Can Claude make my cloned voice speak different languages?

Yes, when the selected ElevenLabs or Cartesia speech model supports the target language. Record the source in your primary language and test a short translated line before generating a longer multilingual voiceover.

Can I replace or delete the cloned voice?

Yes. Claude can replace the source voice on an existing CreativeClaw Character or delete the Character. Replacing it invalidates the old provider copies, while deleting it removes the provider copies and private samples associated with that Character.