The OpenAI Realtime STS integration enables AVR to leverage OpenAI’s real-time speech-to-speech API.
This mode provides low-latency voice conversations, with the added benefit of function calling support for advanced integrations.
For more details on function calls, see AVR Function Calls.
| Variable | Description | Example Value |
|---|---|---|
| PORT | Port on which the STS service runs | 6030 |
| OPENAI_API_KEY | Your OpenAI API key | sk-xxxxxx |
| OPENAI_MODEL | OpenAI model ID to use | gpt-4o-realtime-preview |
| OPENAI_INSTRUCTIONS | # Method 1: Direct variable | "You are a helpful assistant." |
| *OPENAI_URL_INSTRUCTIONS | # Method 2: Web service | https://your-api.com/instructions |
| *OPENAI_FILE_INSTRUCTIONS | # Method 3: Local file | ./instructions.txt |
| OPENAI_TEMPERATURE | Controls randomness in responses (0.0–1.0, default 0.8) | 0.8 |
| OPENAI_MAX_TOKENS | Maximum response length (default: unlimited) | 100 |
avr-sts-openai:
image: agentvoiceresponse/avr-sts-openai
platform: linux/x86_64
container_name: avr-sts-openai
restart: always
environment:
- PORT=6030
- OPENAI_API_KEY=$OPENAI_API_KEY
- OPENAI_MODEL=gpt-4o-realtime-preview
- OPENAI_INSTRUCTIONS="You are a helpful assistant."
- OPENAI_TEMPERATURE=0.8
- OPENAI_MAX_TOKENS=100
networks:
- avr
Update your avr-core service to point to the OpenAI STS endpoint:
avr-core:
image: agentvoiceresponse/avr-core
platform: linux/x86_64
container_name: avr-core
restart: always
environment:
- PORT=5001
- STS_URL=ws://avr-sts-openai:6030
ports:
- 5001:5001
networks:
- avr
The application supports three different methods for loading AI instructions, with a specific priority order:
Set the OPENAI_INSTRUCTIONS environment variable with your custom instructions:
OPENAI_INSTRUCTIONS="You are a specialized customer service agent for a tech company. Always be polite and helpful."
If no environment variable is set, the application can fetch instructions from a web service using the OPENAI_URL_INSTRUCTIONS environment variable:
OPENAI_URL_INSTRUCTIONS=https://your-api.com/instructions
The web service should return a JSON response with a system field containing the instructions:
{
"system": "You are a helpful assistant that provides technical support."
}
The application will include the session UUID in the request headers as X-AVR-UUID for personalized instructions.
If neither environment variable nor web service is configured, the application can load instructions from a local file using the OPENAI_FILE_INSTRUCTIONS environment variable:
OPENAI_FILE_INSTRUCTIONS=./instructions.txt
The file should contain plain text instructions that will be used as the system prompt.
For a complete example, check the docker-compose-openai-realtime.yml in the avr-infra repository.