Exploring Efficient Neural Architectures for Linguistic-Acoustic Mapping in TTS

Santiago Pascual¹, Joan Serrà², Antonio Bonafonte¹

¹Universitat Politècnica de Catalunya, Barcelona, Spain

²Telefónica Research, Barcelona, Spain

This page shows qualitative results for our work on "Exploring Efficient Neural Architectures for Linguistic-Acoustic Mapping in Text-to-Speech". In this work two different pseudo-recurrent mechanisms (QLAD and SALAD) are explored to make acoustic modeling in text-to-speech more efficient whilst trying to maintain the generated speech naturalness.

The code of this project is publicly available in GitHub.