The aim of the project is to create personal synthetic voices for patients that are about to lose their ability to speak as a result of diseases such as Motor Neurone Disease or Throat Cancer.

The Language Technologies Unit has created a platform that allows patients with the aid of speech therapists to create their own personal authentic voice in Welsh and in English. Usually patients will be referred by their speech therapists to the service, with the patients signing a consent form to enable the project to store their voice and create a synthetic voice for them.

The process is web based and patients need to record text provided for them in the form of recording prompts in each language first. They will then be able to generate their own synthetic voice.

In order to communicate orally in Welsh/English, the user types in the web interface and the interface speaks the words aloud with a synthetic version of the patient’s own voice.

A link to the project website may be found here.

This project is a partnership between the NHS and the Language Technologies Unit, Canolfan Bedwyr which is funded under Welsh Governments Cymraeg 2050 Grant Scheme and ran until March 31 2018 as a pilot. As of October 2020 you can now create a Bilingual voice on the Lleisiwr platform rather than a monolingual Welsh and English Voice. Further development work was undertaken in 2021.

For Software Developers the project code for the Lleisiwr project that allows the user to generate and use their digital synthetic voices is available below :-

a)The Voice collection and Banking happens through our modified fork of Mozilla’s Common Voice Project and is integrated into a wordpress page. License

b)Enhancements to MaryTTS( License )and hosting through our Docker Containerisation (License)  that enable voice generation and playback

In the meantime – if you wish to be involved or would like to know more about this project please do not hesitate to get in contact with Stefano Ghazzali (s.ghazzali@bangor.ac.uk)