The AVR App is the official administration panel for Agent Voice Response.
It allows you to design, configure, deploy, and manage AI voice agents connected to Asterisk using a modern web interface.
The AVR App stack includes:
All services are orchestrated via Docker Compose.
Clone the repository:
git clone https://github.com/agentvoiceresponse/avr-infra.git
cd avr-infra
Start the stack:
docker compose -f docker-compose-app.yml up -d
| Service | URL |
|---|---|
| AVR App | http://avr.localhost |
| Backend API | http://api.localhost |
| Traefik Dashboard | http://localhost:8080 |
Open:
Default credentials:
Pipeline agents (ASR + LLM + TTS) use a Provider record per speech-to-text connector. Credentials and tuning live in Provider → config.env (saved per provider in the database). Do not put connector secrets in backend/.env — that file is only for platform services (API URL, Docker socket, Asterisk paths, admin seed).
Workflow: Create Provider (pick ASR template) → Create Agent (pipeline mode: select ASR, LLM, and TTS providers) → assign a Number → run the agent.
| Docker image | Description | Wiki |
|---|---|---|
agentvoiceresponse/avr-asr-deepgram |
Deepgram streaming speech-to-text | Deepgram |
agentvoiceresponse/avr-asr-sarvam |
Sarvam streaming STT (Indic languages) | Sarvam |
agentvoiceresponse/avr-asr-soniox |
Soniox realtime speech recognition | Soniox ASR |
For env key parity across connectors, backend contracts, and admin templates, see backend/docs/AVR-290-asr-env-parity.md in the avr-app repository.
Speech-to-Speech (STS) agents use a Provider record in the admin panel. Credentials and tuning live in Provider → config.env (saved per provider in the database). Do not put connector secrets in backend/.env — that file is only for platform services (API URL, Docker socket, Asterisk paths, admin seed).
Workflow: Create Provider (pick STS template) → Create Agent (STS mode, select provider) → assign a Number → run the agent.
| Docker image | Description | Wiki |
|---|---|---|
agentvoiceresponse/avr-sts-openai |
OpenAI GA Realtime speech-to-speech | OpenAI Realtime STS |
agentvoiceresponse/avr-sts-elevenlabs |
ElevenLabs conversational agent STS | ElevenLabs STS |
agentvoiceresponse/avr-sts-gemini |
Google Gemini Live STS | Gemini STS |
agentvoiceresponse/avr-sts-ultravox |
Ultravox.ai agent or generic-call STS | STS integration |
agentvoiceresponse/avr-sts-deepgram |
Deepgram Voice Agent (ASR + LLM + TTS) | Deepgram |
agentvoiceresponse/avr-sts-speechmatics |
Speechmatics Flow realtime STS | Speechmatics |
agentvoiceresponse/avr-sts-humeai |
HumeAI EVI emotional voice STS | HumeAI STS |
For env key parity across connectors, backend contracts, and admin templates, see backend/docs/AVR-281-sts-env-parity.md in the avr-app repository.
https://demo.app.agentvoiceresponse.com/login
Viewer account:
The AVR App provides a production-like environment locally and is the foundation for scalable conversational AI voice systems.