NaturalSpeech-3

NaturalSpeech3 is a pre-trained foundation text-to-speech model used to evaluate and enhance far-field speaker verification systems by converting near-field speech into far-field speech through adaptive data augmentation.

Papers using NaturalSpeech-3 (3)

Adaptive Data Augmentation With Naturalspeech3 For Far-field Speaker Verification2025 · 1 citesAdaptive Data Augmentation with NaturalSpeech3 for Far-field Speaker Verification2025 · 1 citesE2 TTS: Embarrassingly Easy Fully Non-Autoregressive Zero-Shot TTS2024