First, you need an API key from ElevenLabs.

1. Go to ElevenLabs and sign up (it's free)
2. Once you're in, click your profile icon (top right)
3. Go to "Profile + API Key"
4. Copy your API key

Don't share this key. It's like a password. Keep it safe.

---

Now you need to tell Clawdbot about your ElevenLabs account.

Open your terminal and run:

export ELEVENLABS_API_KEY="your_api_key_here"

Replace your_api_key_here with the key you just copied.

Make it permanent by adding it to your shell config:

echo 'export ELEVENLABS_API_KEY="your_api_key_here"' >> ~/.zshrc
source ~/.zshrc

(If you use bash instead of zsh, replace ~/.zshrc with ~/.bashrc)

---

Edit your Clawdbot config file:

nano ~/.clawdbot/clawdbot.json

Find the section for your agent (probably called "main"), and add:

{
  "agents": {
    "list": [
      {
        "id": "main",
        "tts": {
          "provider": "elevenlabs",
          "voice": "Rachel",
          "enabled": true
        }
      }
    ]
  }
}

Available voices:

• Rachel (default, clear and professional)
• Adam (deeper, more formal)
• Domi (warm, friendly)
• Elli (energetic, younger)
• And 20+ more on ElevenLabs

Pick whichever you like. You can change it later.

Save and exit (Ctrl+X, then Y, then Enter).

---

For the changes to take effect, restart your Clawdbot gateway:

clawdbot gateway restart

Wait about 10 seconds. You should see:

✅ Gateway restarted successfully

---

Now send a message to your Clawdbot and ask it to respond with voice:

You: "Hey, test your voice for me"

If it's working, Clawdbot will reply with both text AND an audio file.

On supported platforms (iMessage, Telegram, Discord), the audio plays automatically. On others, you'll get a download link.

---

ElevenLabs lets you tweak a bunch of settings. Here's what I changed:

Voice Stability

Higher = more consistent but robotic
Lower = more expressive but can sound weird

I like mine at 0.5 (balanced).

Add to your config:

"tts": {
  "provider": "elevenlabs",
  "voice": "Rachel",
  "stability": 0.5,
  "similarity_boost": 0.75
}

Custom Voices

You can clone your own voice or create a custom one in ElevenLabs.

1. Go to ElevenLabs Voice Lab
2. Click "Add Voice"
3. Upload audio samples or use instant voice cloning
4. Copy the voice ID
5. Use it in your Clawdbot config:

"voice": "your_custom_voice_id_here"

---

"Audio file not playing"

Check your channel supports audio. iMessage, Telegram, and Discord work out of the box. If you're using Slack or another platform, you might need to enable attachments.

"API key error"

Make sure you:

1. Copied the full key (no spaces)
2. Added it to your environment variables
3. Restarted Clawdbot after adding it

Run this to verify:

echo $ELEVENLABS_API_KEY

You should see your key. If it's blank, the export didn't work.

"Voice sounds weird"

Try a different voice. Some voices work better for certain types of responses. I use:

• Rachel for general stuff
• Adam when I need a more serious tone
• Domi for casual conversations

---

ElevenLabs pricing (as of Feb 2026):

• Free tier: 10,000 characters/month (~5-10 voice responses depending on length)
• Starter ($5/mo): 30,000 characters/month
• Creator ($22/mo): 100,000 characters/month

For most people, the free tier is enough to start. I upgraded to Starter because I use voice responses a lot.

Pro tip: You can set Clawdbot to only use voice for specific commands. Add this to your config:

"tts": {
  "enabled": true,
  "triggerWords": ["read this", "tell me", "voice"]
}

Now it only speaks when you say one of those phrases. Saves your character quota.

---

I tested a bunch of TTS providers. Here's why ElevenLabs won:

1. Sounds human - Not robotic at all. The first time I heard it, I thought it was a real recording.
2. Fast - Responses come back in 1-2 seconds. Some other providers take 5-10 seconds.
3. Customizable - You can clone voices, adjust tone, add emphasis.
4. Works with Clawdbot - Zero config headaches. Just add the API key and go.

Other options (Google TTS, Amazon Polly) work, but they sound more robotic. If you want your AI to sound like an actual person, ElevenLabs is the move.

Try ElevenLabs free here - no credit card required to start.

---

Once you've got voice working, you can:

• Add wake words - Make Clawdbot listen for "Hey Carlos" and respond with voice automatically
• Voice-to-text input - Talk to Clawdbot instead of typing (requires Whisper API setup)
• Custom voice personas - Create different voices for different tasks (serious voice for work, casual for personal stuff)

I'll write tutorials for those soon. For now, getting basic TTS working is the first step.

---

Q: Can I use this on mobile?
A: Yes. If you're running Clawdbot on a server and messaging via iMessage/Telegram, voice responses work on your phone.

Q: Does this work for all languages?
A: ElevenLabs supports 29 languages. Clawdbot can respond in any of them with the right voice.

Q: Can I use my own voice?
A: Yep. Use ElevenLabs Voice Cloning. Upload 1-5 minutes of your voice, and it'll clone it.

Q: What if I run out of free characters?
A: Either upgrade to a paid plan or set trigger words (see above) so voice only activates when you need it.

---

Adding voice to Clawdbot takes like 10 minutes and makes the whole experience way better. Instead of reading text, you get actual spoken responses that sound human.

If you're using Clawdbot for anything beyond basic commands, I'd 100% recommend setting this up.

Start with ElevenLabs free tier, test it out, and upgrade if you like it.

---

More Clawdbot guides:

• How to Set Up Clawdbot on Mac
• Best AI Coding Tools for Entrepreneurs
• Build Your App Idea in 24 Hours

---

Disclosure: Some links in this article are affiliate links. If you sign up through them, we may earn a commission at no extra cost to you. This helps us keep creating free guides like this one.