B.6 EvaluationMetrics We use two traditional metrics, MSE (or PSNR) and SSIM [51], and a recently proposed deeplearning-based metric LPIPS [52], which measures the similarity between features from different

Deep Neural Networks and Reinforcement Learning methods have empirically shown great promise in tackling challenging combinatorial problems.

In particular, V oicebox outperforms the state-of-the-art zero-shot TTS model V ALL-E on both intelligibility (5.9% vs 1.9% word error rates) and audio similarity (0.580 vs 0.681) while being up to 20 times faster.

In particular, V oicebox outperforms the state-of-the-art zero-shot TTS model V ALL-E on both intelligibility (5.9% vs 1.9% word error rates) and audio similarity (0.580 vs 0.681) while being up to 20 times faster.