Nós-TTS: aWeb User Interface for Galician Text-to-Speech

  1. Carmen Magariños 1
  2. Alp Öktem 3
  3. Antonio Moscoso Sánchez 1
  4. Marta Vázquez Abuín 1
  5. Noelia García Díaz 1
  6. Adina Ioana Vladu 1
  7. Elisa Fernández Rei 1
  8. María Baqueiro Vidal 2
  1. 1 Instituto da Lingua Galega (ILG), Universidade de Santiago de Compostela, Spain
  2. 2 Centro Singular de Investigación en Tecnoloxías Intelixentes (CiTIUS),
  3. 3 Col·lectivaT, Spain
Mostrar afiliacións +
Actas:
Proceedings of the 16th International Conference on Computational Processing of Portuguese - Vol. 2

Ano de publicación: 2024

Páxinas: 200-203

Tipo: Achega congreso

GOOGLE SCHOLAR lock_openAcceso aberto editor

Resumo

Speech synthesis, also known as text-to-speech(TTS), aims to generate human-like speechfrom text. The recent emergence of end-toend deep learning TTS models has led to impressive natural-sounding results. Nevertheless,expanding these models to multiple languagesand speakers poses challenges, especially forlow- or limited-resource languages. In this context, we introduce Nós-TTS, a user-friendlyweb interface for Galician TTS developed under the Nós project. The proposed interfaceoffers a choice among three distinct voicestrained on diverse conditions regarding dataquantity, training approach, and input modality.Although in an experimental stage, informallistening tests have shown a satisfactory performance of the models.