Vocode Setup for GenAI Voice Calls: A Guide for Non-Python Developers

#genai #ai #vocode #voiceai

Target Audience: This guide is designed for developers who are not primarily familiar with Python or its ecosystem, and want to set up Vocode for GenAI voice calls.

Introduction to Vocode and Project Tools
Project Setup and Dependency Management
Configuring Environment Variables and Ngrok
Setting Up the Telephony Server
Creating the Outbound Call Script
Additional Notes and Troubleshooting
References

1. Introduction to Vocode and Project Tools

Vocode is a powerful tool for creating AI-powered voice applications. To streamline the development process, we'll use:

Poetry: For managing project dependencies and virtual environments.
Pyproject.toml: A configuration file used by Poetry to define project details, Python version, and required libraries.

2. Project Setup and Dependency Management

Install Poetry: Follow the instructions on the Poetry installation page.
Create the Project:
```
poetry new project-name
```

Activate the Virtual Environment:

poetry config virtualenvs.create true
poetry shell

Specify the Python Version:
- Open pyproject.toml in your project directory.
- Add the following under [tool.poetry.dependencies]:
```
python = ">=3.9,<3.12"
```
- Update the lock file:
```
poetry lock --no-update
```

Install Dependencies:

poetry add redis^4.5.4 twilio^8.1.0 vocode==0.1.111 vonage^3.5.1 python-dotenv^1.0.0

Verify Installation:
- Create main.py and add:
```
print("Hello, GenAI")
```
- Run:
```
poetry run python main.py
```
- You should see "Hello, GenAI" printed in the console.

3. Configuring Environment Variables and Ngrok

Create the .env File:

Create a .env file in your project's root directory.

Add your credentials (replace placeholders with actual values):

BASE_URL=
DEEPGRAM_API_KEY=
OPENAI_API_KEY=
AZURE_SPEECH_KEY=
AZURE_SPEECH_REGION=
TWILIO_ACCOUNT_SID=
TWILIO_AUTH_TOKEN=
TWILIO_FROM_NUMBER=

Set Up Ngrok:
- Install Ngrok from https://ngrok.com/download.
- Set up your Ngrok AuthToken from https://dashboard.ngrok.com/get-started/your-authtoken.
- Run in a separate terminal:
```
ngrok http 3000
```
- Copy the provided ngrok URL (e.g., https://abcdef123456.ngrok.io) and paste it as your BASE_URL in .env, removing the https:// prefix and trailing slash.
- Image of ngrok terminal output

4. Setting Up the Telephony Server

Copy the Server Code: Get the code from this Gist: https://gist.github.com/bajajcodes/5722cade50a9867b98b246a2cb30ced4 and paste it into main.py.
Activate Your Virtual Environment:
```
poetry shell
```
Run the Server:
```
poetry run python main.py
```

Start the FastAPI Server (Add to main.py):

import uvicorn

if __name__ == "__main__":
uvicorn.run(app, host="localhost", port=port)

Run the Script Again:
```
poetry run python main.py
```

Image of Uvicorn Server Output
Image of ngrok accessing localhost server

5. Creating the Outbound Call Script

Set Up Redis:

docker run -dp 6379:6379 -it redis/redis-stack:latest

Create outbound_call.py: Copy the code from this Gist: https://gist.github.com/bajajcodes/718b598b26e146c4b6bbc784c1d2b1c0 and paste it into outbound_call.py.
Activate Your Virtual Environment:
```
poetry shell
```
Run the Script:
```
poetry run python outbound_call.py
```

Image of Script Console Output
Image of Server Logs

6. Additional Notes and Troubleshooting

Ngrok URL: Ensure the BASE_URL in .env matches the ngrok URL.
Endpoints: Access your FastAPI app at the ngrok URL or your BASE_URL. You can test endpoints like /inbound_call, /events, and /recordings using curl or a web browser.