MODEL DEMOS
PromptTTSD
Controllable Text-to-Spoken Dialogue Synthesis
with Open-Ended Text Descriptions
We trained PromptTTSD by fine-tuning SoulX-Podcast on the PromptDialog dataset. To the best of our knowledge, PromptTTSD is the first open-source spoken dialogue synthesis model to support both zero-shot voice cloning and natural-language style control.
Loading audio demos…