PD-GS fuses ASR-aligned phoneme tokens with HuBERT audio features via a learned gate in a 3DGS talker, improving lip landmark distance (LMD 2.66 on HDTF) over audio-only baselines.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
PD-GS: Phoneme-Driven 3DGS for Audio-Driven Talking Heads
PD-GS fuses ASR-aligned phoneme tokens with HuBERT audio features via a learned gate in a 3DGS talker, improving lip landmark distance (LMD 2.66 on HDTF) over audio-only baselines.