A dependency-free (pure standard-library) Python tool that reads NO-REFERENCE (blind) video-quality scores honestly for the live and UGC cases. It does four things and never decodes video: (1) MAPS a raw NR-model output onto the 1-5 MOS scale with the ITU-T P.1401 monotonic logistic and attaches a confidence BAND = z*RMSE from the model's validation RMSE; (2) A/B COMPARES two clips and REFUSES to declare a winner when the bands overlap or the difference is inside z*RMSE*sqrt(2) (the no-reference
Download free PDF
The handbook · 7 volumes
Fora Soft Video Engineering Handbook
This no-reference score readout tool maps a blind model's output onto the MOS scale with a confidence band and refuses to call an A/B winner when the bands overlap. Our seven-volume handbook covers the video stack; start with Volume 1 on quality metrics and compression artifacts, then Volume 3 on QoE monitoring and Volume 6 on AI models. Written by Nikolay Sapunov, CEO at Fora Soft.
Reading no-reference quality scores for live or UGC video?
Send us your no-reference scores and a few clips, and we will set confidence bands and alert thresholds your team can trust. 250+ video products shipped since 2005.
You get a realistic estimate, technical recommendations and next steps. Free, no obligations.
