Skip to main content
EURASIP Journal on Audio, Speech, and Music Processing

Table 5 Results of AB preference test on vocoders trained with original (NV-MN) and augmented (NV-DA) data

From: Text-to-speech system for low-resource language using cross-lingual transfer learning and data augmentation

NV-MN (baseline)

NV-DA

No preference

21.61%

16.09%

62.30%

Back to article page