ViSQOL (Virtual Speech Quality Objective Listener) is an open, full-reference quality metric developed at Google that estimates perceived audio quality by measuring the spectro-temporal similarity between a reference and a degraded signal, mapping it to a MOS-like score. Unlike the telephony-focused PESQ and POLQA, ViSQOL has modes for both speech and general audio, making it useful for evaluating music codecs and neural codecs as well as voice. Being free and open, it became a popular way for researchers and engineers to benchmark codecs and enhancement systems quickly, without licensing costs or human listening panels. Like all objective metrics it is a proxy for subjective MOS, best used for relative comparison and trend-spotting rather than absolute verdicts.
The book · Volume 5 of 7
Audio for Video: A Complete Guide to Sound in Video Products: From Jitter Buffers and Lip Sync to Atmos, Quality Metrics, and Neural Codecs
ViSQOL lets a team score codec and network changes without booking a listening panel, as long as they know what the number hides. Volume 5 covers ViSQOL in detail alongside PESQ, POLQA and MUSHRA, plus the packet loss concealment, FEC and neural codec behaviour these metrics are used to judge. Written by Nikolay Sapunov, CEO at Fora Soft.
Need an objective audio quality score like ViSQOL in your release pipeline?
We will set up automated ViSQOL scoring for your codec or network changes and give you baseline scores and pass thresholds you can run on every build. 250+ video products shipped since 2005.
