SOMOS

The SOMOS dataset is an open-access benchmark used to evaluate mean opinion score (MOS) prediction models in the context of text-to-speech systems for low-resource languages.

Papers using SOMOS (4)

Neural networks for Text-to-Speech evaluation2026Resource-Efficient Fine-Tuning Strategies for Automatic MOS Prediction in Text-to-Speech for Low-Resource Languages2023 · 2 citesEnabling Auditory Large Language Models for Automatic Speech Quality Evaluation2024 · 1 citesSOMOS: The Samsung Open MOS Dataset for the Evaluation of Neural Text-to-Speech Synthesis2022