표 1.|Table 1. TTS와 실제 음성의 MOS|MOS of synthesized speech & ground truth speech

Method	MOS (신뢰구간 95%)
Tacotron 2 - WaveNet	3.86±0.36
Research model	3.8±0.12
Ground truth	4.6±0.07

MOS, Mean Opinion Score; TTS, Text-to-Speech.