Rectified flows with a tunable sampling temperature offer the best naturalness-diversity trade-off among stochastic prosody predictors for text-to-speech.
The models were trained for 100k steps with a batch size of 32, which allowed all models to converge
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
-
Investigating Stochastic Methods for Prosody Modeling in Speech Synthesis
Rectified flows with a tunable sampling temperature offer the best naturalness-diversity trade-off among stochastic prosody predictors for text-to-speech.