Festival

Festival is a multi-lingual speech synthesis system providing a full framework for building concatenative and unit-selection voices.

Developed at the University of Edinburgh's Centre for Speech Technology Research (CSTR), Festival offers a complete architecture for text-to-speech (TTS) research and deployment. It supports diverse languages including English, Spanish, and Welsh through a modular C++ core and a Scheme-based command interpreter. The system integrates the Edinburgh Speech Tools library to handle signal processing tasks (like pitch tracking and LPC analysis) while allowing developers to implement custom prosody models and linguistic front-ends. It remains a foundational tool for academic researchers and developers needing a transparent, highly configurable engine for voice synthesis.

https://www.cstr.ed.ac.uk/projects/festival/