Advanced
Integrations
How to connect Burki Voice AI with Twilio, Deepgram, OpenAI, ElevenLabs, and more.
Burki Voice AI integrates seamlessly with leading telephony, speech, and AI providers. This page explains what each integration does and how to configure it.
For troubleshooting integration issues, see the FAQ or open an issue on GitHub.
Integration Flow
Twilio (Telephony)
Twilio (Telephony)
Purpose: Handles phone calls, media streaming, and SMS.Setup:
- Sign up at Twilio.
- Buy a phone number and enable Media Streams.
- Set environment variables:
TWILIO_ACCOUNT_SIDTWILIO_AUTH_TOKEN
- Configure Twilio webhook to point to your Burki backend.
Speech-to-Text (STT) Providers
Speech-to-Text (STT) Providers
Deepgram (default, recommended)
- Real-time transcription of audio calls.
- Set
DEEPGRAM_API_KEYin your.env.
- You can extend Burki to use other STT APIs by customizing the backend.
Language Model (LLM) Providers
Language Model (LLM) Providers
OpenAI (default, recommended)
- Generates intelligent responses using GPT models.
- Set
OPENAI_API_KEYin your.env.
- Use
INWORLD_BEARER_TOKENfor Inworld LLM.
- Configure provider and API key in assistant settings or
.env.
Text-to-Speech (TTS) Providers
Text-to-Speech (TTS) Providers
ElevenLabs (default, recommended)
- Converts text responses to natural-sounding speech.
- Set
ELEVENLABS_API_KEYin your.env.
- Use Deepgram for TTS if enabled in your assistant config.
- Use Inworld for TTS with
INWORLD_BEARER_TOKEN.
- Use
RESEMBLE_API_KEYfor Resemble TTS.
Storage (S3, etc.)
Storage (S3, etc.)
- Stores call recordings and large audio files.
- Set S3 environment variables if using cloud storage.
Database (PostgreSQL)
Database (PostgreSQL)
- Stores all app data (users, calls, assistants, etc.)
- Set
DATABASE_URLin your.envfile.