Glossary
MOS (mean opinion score)
The standard subjective measure of speech and audio quality, in which listeners rate samples from 1 (bad) to 5 (excellent) and the ratings are averaged.
Protocols such as ITU-T P.800, and its crowdsourced counterpart P.808, define how the listening tests are run. Natural human recordings typically rate in the mid-4s, and strong modern TTS systems approach them. Because MOS is subjective, scores only compare within the same test and listener pool; automatic predictors exist but are calibrated against human MOS.
Why it matters
TTS quality claims are made in MOS, so the cleanliness and consistency of voice data feed directly into it.