Recommended OS and HW (PSP)

Recommended (*) Operating Systems:

• • Windows 64-bit – latest version of Windows 10, Windows Server 2019
  • Linux 64-bit – latest version of RHEL/CentOS 7 

Compatible (**) Operating Systems:

• • 64-bit Windows 7, Windows 8.1, Windows Server 2016, …
  • 64-bit Linux with glibc >= 2.17, e.g. Ubuntu 20.04, Mint 19.3, RHEL/CentOS 8.2, … 

(*) Speech Platform components (e.g. Speech Engine) is tested internally on these systems. 

(**) Speech Platform components (e.g. Speech Engine) is known to be successfully deployed on these systems. 

Recommended Hardware

Required HW is linked to a set of technologies (the SPE configuration) and the load that should be processed per day (or during a peak hour). It is highly recommended to run a load test with selected configuration and evaluation licenses before purchasing HW. Here are recommendations for typical configurations:

Speech Platform – Voice Biometrics configuration – 100 hours/day 

• • CPU: 8 physical cores, typically 1x Intel® Xeon E5-2640 v4
  • RAM: 32 GB
  • HDD: 100 GB (depends on the archive policy)
  • Configuration includes: SID4_L4, LID_L4, GID_L4, AGE_L4, VAD, SQE 

 Speech Platform – Transcription System configuration – 100 hours/day 

• • CPU: 8 physical cores, typically 1x Intel® Xeon E5-2640 v4
  • RAM: 32 GB
  • HDD: 100 GB (depends on the archive policy)
  • Configuration includes: STT 5th generation – 2 languages (half load each), KWS 5th generation – 2 languages, LID_L4, VAD, SQE  

Speech Platform – Voice Biometrics and Transcription System configuration – 100 hours/day 

• • CPU: 14 physical cores, typically 1x Intel® Xeon Gold 5120;
  • RAM: 64 GB
  • HDD: 500 GB (depends on the archive policy)
  • Configuration includes: SID4_L4, LID_L4, GID_L4, AGE_L4, STT 5th generation – 2 languages (half load each), KWS 5th generation – 2 languages, VAD, SQE