This repository collects small AI experiments, reference code, and architecture documents focused on realtime voice integration (LiveKit, WebRTC) and voice model flows (ElevenLabs). It's intended as a lightweight playground and documentation hub you can fork, extend, and use to prototype integrations quickly.
src/
— minimal Python starter CLI and example code. Use it as the base for experiment scripts.docs/
— Markdown architecture docs (Mermaid.js diagrams) and any rendered artifacts.scripts/
— helper scripts for environment setup (PowerShell helpers to install Python + venv and to install repo-local git hooks)..githooks/
— local git hook scripts (install locally with the installer script inscripts/
).requirements.txt
— Python dependencies for the starter and small utilities.
docs/livekit-architecture.md
— LiveKit + ElevenLabs voice flow and component responsibilities.docs/webrtc-architecture.md
— WebRTC session-level overview, signaling and ICE responsibilities.docs/elevenlabs-v3-architecture.md
— ElevenLabs TTS/embedding flow, API considerations and example sequences.
This guide provides the steps to set up and run the local "Jarvis" AI agent on Windows using PowerShell, as specified in the project's README.md
[1, 2].
- Windows with PowerShell.
- Git installed.
- Your ElevenLabs API key.
If you haven't already, clone the artificial-intelligence
repository to your local machine.
git clone https://github.com/Archit3115/artificial-intelligence.git
cd artificial-intelligence
This project includes a helper script to install a project-local version of Python and create a virtual environment named ai-dev.
Run the following command in PowerShell. You may be prompted for administrative permissions to proceed.
. scripts\setup_python_and_venv.ps1
The agent requires your ElevenLabs API key to generate speech. Set it as an environment variable for your current PowerShell session. Replace "your_key_here"
with your actual key.
$env:ELEVEN_API_KEY="your_key_here"
Note: This variable is only set for the current terminal session. You will need to set it again if you open a new terminal.
Activate the virtual environment and install the required Python packages from the requirements.txt
file.
The Set-ExecutionPolicy
command may be needed to allow the activation script to run.
Set-ExecutionPolicy -Scope Process -ExecutionPolicy Bypass
. .\ai-dev\Scripts\Activate.ps1
python -m pip install -r requirements.txt
With the environment set up and dependencies installed, you can now run the agent. The main script is located in the src/
directory, which is the designated location for the starter CLI and example code.
python -m src.main
The agent will start, greet you, and begin listening for your commands through your microphone.
This guide is based on the instructions provided in the Archit3115/artificial-intelligence repository.
- Diagrams are stored as Mermaid blocks inside
docs/*.md
. - CI: a GitHub Actions workflow can render these to
docs/diagrams/
(see.github/workflows/render-mermaid.yml
). - If GitHub shows a blank file for a doc you edited locally, ensure you saved the file to the repo path and committed it:
git update-index --no-assume-unchanged docs\livekit-architecture.md
git add docs\livekit-architecture.md
git commit -m "docs: restore livekit-architecture.md"
git push origin dev-101
- Create a branch from
main
ordev-101
for your changes. Open a PR back tomain
when ready. - We try to keep architecture docs simple (one Mermaid fenced block per file) so GitHub's renderer and the CI renderer behave consistently.
- Maintainer: Archit Srivastasva architsrivastava3115@gmail.com
- No license file in this repo. Add
LICENSE
if you intend to make this code public under a formal license.
--
Short and practical — use the docs/
files as architecture references and the src/
starter to prototype quickly.