A phoneme-guided diffusion model, PGDI, reconstructs lost speech segments up to one second long, with or without a transcript, while preserving speaker identity and prosody.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.AS 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Transient Noise Removal via Diffusion-based Speech Inpainting
A phoneme-guided diffusion model, PGDI, reconstructs lost speech segments up to one second long, with or without a transcript, while preserving speaker identity and prosody.