There are three major dialects of Irish: Ulster Irish, Connaught Irish and Munster Irish. Although they share many grammatical features there are notable differences too. The most striking difference is the pronunciation. It's immediately clear which dialect we're dealing with as soon as a native speaker starts to speak. There is no such thing as a received pronunciation in Irish - although there is a written standard.

Thus, when developing a synthesiser for Irish a voice for each one of the (major) dialects has to be created. So far we have three voices: Gweedore (Ulster), Ráth Chairn (Connaught) and the Dingle Peninsula (Munster).

We intend to provide more voices, especially voices for the various sub-dialects (e.g. Mayo Irish or Waterford Irish). We strongly encourage native speakers of any of the major dialects or any of the sub-dialects to contact us.

There are three methods of speech synthesis on this page: HTS, DNN agus Unit Selection. All methods have their own strengths and drawbacks. Here's a short guide.

HTS

A fast synthesis system that is suitable for real time systems. HTS voices are stable and easy to understand although they don't sound quite as natural as the DNN voices.

DNN

This method of synthesis is based on Deep Neural Networks. These voices sound very natural but they take a bit longer to synthesize sentences. They are suitable for education, announcements and newspapers.

Unit Selection

Unit Selection was the first synthesis system we developed. It's well suited for short phrases but the quality of the output depends on the input. We are currently not planning to develop this type of synthesis for new voices. It's only available for Ulster and Connaught Irish.

ABAIR is a project of the Phonetics and Speech Laboratory, Trinity College, Dublin. ABAIR develops synthetic voices and speech recognitions systems for Irish.

Speech Synthesis

There are a lot of use cases for text-to-speech (TTS) synthesis. It is used in public places, e.g. for announcements at train stations or aiports. TTS synthesis is extremely important for the visually impaired, because it enables them to use computers or smart phones. The amount of teaching material that incorporates TTS synthesis is constantly increasing too.

People also like to use TTS synthesis to have newspaper articles read out to them while they are driving or jogging. In fact, TTS synthesis is a must for any language these days. More information.

Speech Recognition

Speech recognition enables us to interact with computers by simply "talking" to them, e.g. when you tell your GPS where to go or when you ask your smart phone for the weather forecast. There's an increasing demand for a speech recognition system for Irish, and ABAIR is currently developing one. More information.