Phone duration prediction and per-phone k-means quantization raise kNN-VC's privacy EER from 10% to nearly 50%, but target-selection changes can erase most of that gain.
As target speakers, we use 100 speakers from the LibriTTS [30]train-other-500dataset, evenly distributed across genders
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
dataset 1
citation-polarity summary
fields
eess.AS 1years
2025 1verdicts
CONDITIONAL 1roles
dataset 1polarities
use dataset 1representative citing papers
citing papers explorer
-
Private kNN-VC: Interpretable Anonymization of Converted Speech
Phone duration prediction and per-phone k-means quantization raise kNN-VC's privacy EER from 10% to nearly 50%, but target-selection changes can erase most of that gain.