Multilingual website translation is the main new feature added in version 3.1 of the open source transLectures-UPV Platform (TLP) for the integration of automated transcription and translation technologies into media repositories, just released by the MLLP research group. You can watch here our new video demonstration to see how it works.
The transLectures-UPV Platform (TLP), developed by the Universitat Politècnica de València’s (UPV) Machine Learning and Language Processing (MLLP) research group, is an open source piece of software designed to integrate automated transcription and translation technologies into media repositories. Its main components are the TLP Database, Web Service, Ingest Service, Text Translation Editor, Media Player, and Website Translation Plugin. The MLLP is releasing TLP version 3.1, now available for download from the TLP page.
With this new release, TLP now provides support for multilingual website translation, in addition to the already existing support for automatic multilingual media subtitling and document translation (full MOOC content support).
TLP 3.1 makes it very simple to add automatically translated versions to a website, making use of statistical models for machine translation. It also provides the ability for inline post-editing of the the translated versions, allowing web content managers to revise the translated texts on the website itself.
As a first implementation of the new website translation features, we have made the MLLP website multilingual, with automatically translated versions in Catalan, Spanish, French, Italian and other languages, in addition to the original English-language version; you can use the language selector at the top right of this website in order to change the language in any section. We have also prepared the video demonstration that you can watch above.
The media upload process has also been improved in TLP 3.1, which now can show estimated processing times according to the length of the files and on the features requested, in order to help users organize their workflows.
A complete description of the functionalities of TLP 3.1 can be found in the TLP documentation.
The fastest way to try the MLLP’s technology is to access the MLLP transcription and translation platform, which integrates our own software TLP and TLK (the transLectures-UPV Toolkit for speech recognition), our machine translation technology, and our text-to-speech synthesis technology. Just open a trial account and start uploading media and text files to try our automatic transcription and translation technology, and even integrate it with your own media repository via our API.
The MLLP research group, made up of researchers based at the Universitat Politècnica de València (UPV), have been developing for years their expertise and tools for machine learning and language processing in the fields of educational technologies and big data within the ideal framework of the EU projects transLectures and EMMA. The MLLP has been an integral part of both projects, and within the ongoing project EMMA, TLP is being intensively used for the full transcription and translation of MOOC contents in several languages from leading European universities.
The MLLP’s TLP and TLK are also being used as the basis for automatic subtitle generation in the educational media repositories of our own Universitat Politècnica de València (Polimedia), and of the Carlos III University of Madrid (you can learn more about our projects here). So far, our technology has been used to generate multilingual subtitles for over 18 000 media files (more than 3 700 hours) and translations for over 8 700 text documents (more than 59 000 sentences).
So download TLP 3.1, or try it now in the MLLP transcription and translation platform. We’ll keep you updated on new developments and releases of our software; just keep checking our website and follow us on Twitter, @mllpresearch!