Skip to main content

System Requirements

  • CPU: 2.6 GHz+, 64-bit. For maximum speed, a CPU with a large L3 cache (15 MB+) is recommended.
  • RAM: 4 GB+ minimum. Memory usage scales with recording length: processing typical recordings (~10 minutes) uses under 1 GB, while processing long recordings (e.g., 3.5 hours) may require up to 13 GB. Size RAM according to the expected length of the audio being processed.
  • Storage: Larger HDD/SSD capacity may be required depending on how data is processed and stored.
  • GPU (optional): Compatible NVIDIA GPU with 6+ GB of VRAM recommended for typical use. VRAM usage scales with recording length (and sample rate): short recordings use under 1.5 GB, while long recordings (~80 min) may require up to ~9 GB. A GPU-accelerated version of Voice Inspector is available to improve the processing speed of Deepfake detection — Hardware Acceleration below.

Supported Operating Systems

Windows

Voice Inspector is supported and tested on Windows 11 (64-bit, x86_64). Due to Windows' backward compatibility, the application is expected to run correctly on Windows 10 as well.

Linux

Voice Inspector is built and validated on Rocky Linux 8 (64-bit, x86_64) runners. As a result, it is compatible with any Linux distribution providing glibc ≥ 2.28.

Required Dependencies

Windows

Voice Inspector requires the Microsoft Visual C++ Redistributable (2015 or later). Most Windows systems already have this installed as part of other software; it typically needs to be installed manually only on a clean or minimal Windows installation. Its absence results in a runtime error referencing a missing MSVCP140.dll. It can be installed with:

winget install --id Microsoft.VCRedist.2015+.x64 -e

Rocky Linux 8

On a fresh Rocky Linux 8 system, the following packages must be installed manually, as they are not included by default. Without them, the application cannot be launched:

sudo dnf install epel-release -y
sudo dnf install xcb-util-cursor libglvnd-opengl -y

Hardware Acceleration

Hardware acceleration requires matching, compatible versions of Voice Inspector (VIN-cuda-*), CUDA 13, and the GPU driver. Using mismatched versions may prevent hardware acceleration from being enabled; the application will still work, falling back to CPU-only processing.