WhatsApp & Telegram Voice Messages with AI: My Setup

When I'm driving, I can't type. But my AI assistant still sends me updates — as voice messages in my own voice. Here's how I set up ElevenLabs for automatic voice messages.

The Vision: Proactive Voice Messages

Imagine your AI assistant automatically informs you about important events — not as text, but as natural voice messages. It's not just more practical while driving, it also feels much more personal.

What my setup does:

• Sends me a daily briefing as a voice message (appointments, weather, important emails)
• Notifies me of urgent events
• Responds to my voice messages with voice messages
• Uses my cloned voice — feels like a conversation with myself

The Components

1. ElevenLabs Account: For text-to-speech synthesis
2. Messaging Integration: Telegram Bot API or WhatsApp Business API
3. Orchestration: n8n, Make, or custom code
4. Optional: Voice clone for personalized voice

Step 1: Set Up the ElevenLabs API

First you need API access at ElevenLabs:

1. Sign up and choose at least the Starter plan ($5/month for 30,000 characters)
2. Go to "Profile" → "API Key" and copy your key
3. Note the Voice ID of your desired voice (found in the Voice Library)

API Test with curl

curl -X POST "https://api.elevenlabs.io/v1/text-to-speech/VOICE_ID" \
     -H "xi-api-key: YOUR_API_KEY" \
     -H "Content-Type: application/json" \
     -d '{"text":"Hello, this is a test.", "model_id":"eleven_multilingual_v2"}' \
     --output test.mp3

Step 2: Create a Telegram Bot

Telegram is easier to set up than WhatsApp:

1. Open Telegram and search for "@BotFather"
2. Send /newbot and follow the instructions
3. You'll receive a bot token — keep it safe

Telegram bots send voice messages as OGG (Opus). ElevenLabs delivers MP3, so you need conversion:

# Convert with ffmpeg
ffmpeg -i input.mp3 -c:a libopus output.ogg

Step 3: Automation with n8n

n8n is an open-source automation tool perfect for this use case.

Workflow Overview

1. Trigger: Scheduled (cron) or event-based (webhook)
2. Generate text: Create the message with OpenAI or your own prompt
3. TTS: Send text to ElevenLabs API, receive audio file
4. Convert: MP3 to OGG for Telegram
5. Send: Voice message via Telegram Bot API

Step 4: WhatsApp Integration (Alternative)

WhatsApp is more complex but often preferred for business:

• Option A: WhatsApp Business API — Official, requires business verification, per-message costs, reliable
• Option B: WhatsApp Web Gateway — Unofficial but easier, tools like Baileys or whatsapp-web.js, risk of account suspension

For personal use, I recommend Telegram — it's developer-friendlier with fewer restrictions.

Step 5: Voice Clone for Personalization

The secret sauce: Generate voice messages in your own voice.

1. Record 3-5 minutes of your voice (clear, no background noise)
2. Upload the sample to ElevenLabs Voice Lab
3. Choose "Professional Voice Cloning" for best quality
4. Use the new Voice ID in your automation

The result: Your AI assistant speaks with your voice. Surreal, but incredibly useful.

Practical Use Cases

Daily Briefing

Every morning at 7:00 AM — today's appointments, important emails, weather forecast.

Urgent Notifications

Real-time alerts for important emails, calendar reminders, monitoring alerts.

Bidirectional Conversation

You send a voice message → Whisper API transcribes → GPT generates response → ElevenLabs converts to speech → Bot sends voice message back.

Cost Calculation

A typical voice message has 200-500 characters. With ElevenLabs:

Conclusion: Your Personal Voice Assistant

With ElevenLabs, a messaging bot, and some automation, you have an assistant that proactively communicates with you in your own voice. This isn't science fiction — I use this setup daily.

Setup takes about 2-4 hours. After that, everything runs automatically. For me, it's changed how I interact with my digital systems — less screen time, more hands-free usage.