> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/nemo/labs-voice-agent/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/nemo/labs-voice-agent/_mcp/server.

# Prerequisites

Verify the requirements for the workflow you plan to run before installing NeMo Voice Agent.

## Hardware Requirements

Verify that your machine and interactive audio setup meet these hardware requirements:

| Requirement | Detail |
| --- | --- |
| GPU | An NVIDIA GPU and driver compatible with the CUDA wheel selected in `pyproject.toml`. The default is CUDA 13.0. |
| Model support | The default NVFP4 language model requires hardware with FP4 support. Speech and text-to-speech models also use CUDA by default. |
| Audio devices | A microphone and speaker available to the browser for an interactive voice session. |

## Software Requirements

Install or make available the software required by the workflow you plan to run:

| Requirement | Detail |
| --- | --- |
| Operating system | Linux. `install.sh` installs system packages with `apt-get`. With another package manager, install the equivalent packages yourself. |
| Python | Python 3.12 or 3.13. The project configures `uv` to download and manage a compatible interpreter. |
| Node.js and npm | Required for the browser client. `install.sh` installs both on systems that provide `apt-get`. |

## Pre-Installation Checklist

Run these commands to confirm that the GPU and optional browser tools are available:

```bash
nvidia-smi
node --version
npm --version
```

## Troubleshoot Prerequisites

Do not run the installer inside a non-`base` conda environment. Run `conda deactivate` first. This prevents
the conda compiler and Python headers from interfering with packages that build native extensions.

## Next Steps

After the prerequisite checks pass, install the project or review the complete first-run workflow:

- [Installation](/nemo/labs-voice-agent/get-started/installation) creates the virtual environment and installs dependencies.
- [Quickstart](/nemo/labs-voice-agent/get-started/quickstart) starts the model server, voice-agent server, and browser client.