Artificial intelligence and language technologies
Elhuyar | Zerbitzuak
Elhuyar and the Orai Artificial Intelligence Center are working together to develop artificial intelligence technology. This alliance allows us to explore cutting-edge research fields, and then transfer the results to the services.
These are the services and products developed and marketed by Elhuyar:
Elia, the automatic translator for Elhuyar
Elhuyar has many years of experience working on machine translation projects. In 2007, for example, he participated with the research team Ixa of the UPV/EHU in the project to create the system called Matxin, a milestone in the automatic translation in Basque, which consisted of an automatic translation system based on free software that translated from Spanish to Basque. It was the first automatic translator of those characteristics.
As research has advanced, other techniques have been incorporated for the creation of automatic translators, which are currently based on artificial intelligence and neural networks. We offer an automatic translation service available to users and customers: Elia. Elia, the automatic translator for Elhuyar, has the following features:
- Speaks 6 languages: Basque, Spanish, French, English, Catalan and Galician.
- Enables full documents to be translated. In addition, it allows the translation proposed by the system to be changed sentence to sentence.
- It offers free applications for both desktop and mobile.
- It offers the possibility to personalize translations: you can be added a custom dictionary to translate certain words or phrases according to the preferences of each user.
- It has a specific system for legal and administrative texts.
- Offers guided translation. Users can thus make changes to the translation proposed by the system, with the help of the system: by clicking on a word, the system will offer different possibilities for the text immediately after that word.
- Integration into professional translators tools: Wordfast, SDL Trados...
- Confidentiality and security: User data and information sent through the service are handled securely.
- Thanks to its REST API, it can be integrated into any application.
For more information please contact:
Development of cross-border cooperation and knowledge transfer in language technologies (EFA 227/16/LINGUATEC)
The project aims to develop, demonstrate and disseminate new resources, tools and innovative linguistic applications that improve the digitisation level of Aragonese, Euskera and Occitan. The Project will allow technological cooperation between languages to develop new language resources and tools, as well as new on-line applications to facilitate the use of the three languages and multilingual access to content by end users. Financing:
|
Aditu, Elhuyar's speech recognizer
Aditu is the result of research conducted with voice recognition technology. It is a platform that allows you to convert the voice of video and audio files into text, and then you can edit the result.
It makes automatic transcriptions of video and audio files in Basque and Spanish, as well as bilingual files. It also allows the creation of subtitles automatically, making it an ideal tool to ensure accessibility in audiovisual products.
Aditu allows the automatic transcription and subtitling of audio-visual in a few minutes, making it a very useful system for subtitling meetings, municipal plenary sessions or documentaries.
These are the main functions offered by Aditu:
- Transcriptions and automatic subtitles of audiovisual files with interventions in Basque and Spanish.
- It has an advanced editing interface for correction of subtitles and transcriptions, which allows the user to easily make the necessary adaptations.
- Allows automatic translation of subtitles and transcripts through the automatic translator Elia.
- It allows the download of subtitles and transcriptions in different formats and, in the case of subtitles, you can also download the video with the subtitles already inserted.
- Confidentiality and security: User data and information sent through the service are handled securely.
- Thanks to its REST API, it can be integrated into any application.
More information on the aditu.eus website or on aditu@elhuyar.eus
Elhuyar neuronal TTS
Elhuyar offers several text-to-speech conversion systems on the website ttsneuronala.elhuyar.eus. These are synthetic voices that sound natural and are able to read any text. Today, it offers voices in Basque and Spanish.
At Elhuyar we have created several TTS voices and made them available to users and customers. On the one hand, we have generated voices that are not identifiable, that is, voices that when heard do not refer us to a concrete person. On the other hand, we have also created systems that mimic the voice of specific people, so we also offer the possibility of creating personalized voices.
TTS custom voices are used in the following Elhuyar services: The Elhuyar dictionary, the Elhuyar magazine and the website zientzia.
There are several ways to use the voices available: through the reading bar, the web service or integrating them through the REST API.
More information on the website ttsneuronala.elhuyar.eus and on tts@elhuyar.eus
More smart solutions
We also develop other artificial intelligence projects and language technologies, such as the automatic extraction of terminology and knowledge, the management and advanced creation of linguistic data, solutions linked to digital humanities and services for smart and advanced language management.
- One of these projects is LINGUATEC IA, a project to advance the digitalization of Aragonese, Catalan, Basque and Occitan, through artificial intelligence. Co-financed by FEDER through the POCTEFA Programme 2021-2027 and led by Elhuyar (through its Orai artificial intelligence centre), it aims to apply new models of generative language in these languages.
- The Deep Learning project for the development of a multi-lingual automatic subtitling and folding system started in October 2022 and ended in May 2024. The project was funded by the European Union – NextGenerationEU.