Audio samples generated by two models A and B
Model A: T5_US_Accent_1
Trained on: LJ Speech Dataset
Description: This model has been fine-tuned to produce a more natural and expressive speech synthesis with a typical American accent.
Model B: T5_US_Accent_2
Trained on: LibriVox Dataset
Description: The second model has a more neutral tone, with a vibe reminiscent of the AI assistant "Friday" from the Iron Man series.
Note: None of these sentences were part of either training set.
For each sample, I've marked the model that sounds best with the text by adding (preferred) next to it.