Multi-speaker and Multi-dialectal Catalan TTS Models for Video Gaming

Recently, we explored and trained different state-of-the-art text-to-speech (TTS) architectures for Catalan. We used existing datasets but also produced a new Catalan multi-accent dataset to train these architectures. The objective of our work is to improve the quality of current TTS systems in Catalan and export the resulting models for potential interactive applications and video games.