The LINGUATEC-AI project, led by the technology center ORAI, needs to build an Aranese recognition system (ASR) to develop a voice synthesizer. To achieve this, at least 175 hours of transcribed audio with a wide diversity of speakers are required. Currently, 8 hours recorded during a voice marathon in 2022 are available, but 167 more hours are needed.
The call is aimed at Aranese residents to contribute their voices, seeking to record the widest possible variety of ways of speaking the native language of the Aran Valley. Participants do not need to have Aranese as their mother tongue. To ensure diversity, the project seeks a balance in gender, ages (from under 18 to over 70), origin from the different regions of the Valley (Naut Aran, Mijaran, Baish Aran), and a mix of native and second-language speakers.
Each speaker can contribute a maximum of 2-3% of the total recordings, equivalent to one and a half hours. Additionally, the recorded texts are requested to cover varied styles and registers, including literary, legal, journalistic, institutional fragments, and informal conversations, selected by the Language Department.
To ensure a consistent format, a professional recording studio will be made available. Interested participants wishing to contribute their voice must complete an online form available until Monday, October 20th, at 11:45 PM.
This initiative is part of the LINGUATEC-AI project, which focuses on implementing artificial intelligence in low-resource languages of the Pyrenees such as Aragonese, Catalan, Basque, and Occitan (including Aranese). The Government of Aran, along with Lo Congrès permanent dera Lengua occitana, is committed to the digitalization and presence of its native language in digital media.
