System Requirements

This configuration is intended for developing and testing the speech-to-text systems based on the Neurotechnology AI SDK, as well as deploying the end-user system with ASR-2 and Diarization-4 components.

• x86-64 (64-bit) processor is required. These processors showed acceptable performance during our internal testings:
  • Intel Core i7-14700K
  • AMD Ryzen 7950X
• AVX2 support is required. Most modern processors support this instruction set, but please check if a particular processor model supports it.
• 32 GB RAM is required for general usage scenarios.
• Network/LAN connection (TCP/IP) for client/server applications.
• Microsoft Windows specific:
  • Microsoft Windows 10 / 11.
  • Microsoft .NET framework 4.8 (for .NET components usage)
  • Microsoft Visual Studio 2022 or newer (for application development with C++ / C# / VB .NET)
  • Java SE JDK 11 or newer (for application development with Java)
  • Python 3.x (for application development with Python)
• Linux specific:
  • Linux 4.19 or newer kernel (64-bit) is required.
  • At least Ubuntu 20.04 or Debian 11 or RHEL 8 is recommended
  • glibc 2.28 or newer
  • libgudev-1.0 232 or newer (for microphone usage)
  • alsa-lib 1.1.8 or newer (for voice capture)
  • gcc 8.3 or newer (for application development)
  • GNU Make 3.81 or newer (for application development)
  • Java SE JDK 11 or newer (for application development with Java)
  • Python 3.x (for application development with Python)

This configuration with a single GPU is intended for developing and testing the speech-to-text systems based on the Neurotechnology AI SDK, as well as deploying the end-user system with ASR-10 and Diarization-20 or Diarization-60 components. A similar configuration with two GPU units can be used for deploying a higher performance system with ASR-30 component

• x86-64 (64-bit) processor is required.
• AVX2 support is required. Most modern processors support this instruction set, but please check if a particular processor model supports it.
• 32 GB RAM is required for general usage scenarios.
  • 64 GB recommended for multi-worker deployments.
• GPU-specific requirements:
  • NVIDIA GPU with Compute Capacity 7.5 or better
    • Single GeForce RTX 2080 unit is enough for using ASR-10 and Diarization-20 or Diarization-60 components
    • Two GeForce RTX 2080 units or newer comparable GPU(s) are needed for using ASR-30 component
  • 4 GB VRAM for each ASR engine worker deployment
  • 1 GB VRAM for each Diarization engine worker deployment
  • 20-30% of free VRAM needed for allocator headroom
• Network/LAN connection (TCP/IP) for client/server applications.
• Microsoft Windows specific:
  • Microsoft Windows 10 / 11.
  • NVIDIA driver 570.65 or newer
  • Microsoft .NET framework 4.8 (for .NET components usage)
  • Microsoft Visual Studio 2022 or newer (for application development with C++ / C# / VB .NET)
  • Java SE JDK 11 or newer (for application development with Java)
  • Python 3.x (for application development with Python)
• Linux specific:
  • Linux 4.19 or newer kernel (64-bit) is required.
  • At least Ubuntu 20.04 or Debian 11 or RHEL 8 is recommended
  • NVIDIA driver 570.26 or newer
  • glibc 2.28 or newer
  • libgudev-1.0 232 or newer (for microphone usage)
  • alsa-lib 1.1.8 or newer (for voice capture)
  • gcc 8.3 or newer (for application development)
  • GNU Make 3.81 or newer (for application development)
  • Java SE JDK 11 or newer (for application development with Java)
  • Python 3.x (for application development with Python)

The configuration with multiple GPU is intended for deploying a high-performance system with ASR-100 and Diarization-200 components. This hardware was used during internal testings and allowed to achieve the specified performance.

• 2x AMD EPYC 9334 32-Core Processor
• 768 GB RAM
• 2x 1.92TB NVME SSD
• 8x NVIDIA RTX PRO 6000 96GB PCIe
• SuperMicro 420GP-TNR Server
• Ubuntu Server 24.04 OS

Please contact us to find out if your hardware setup would be suitable in this scenario.