TLK v. 1.3.0 now available for download

Just a quick announcement before the summer to let you know that the transLectures-UPV toolkit (TLK) has been updated and is now available for download.

Version 1.3.0 of the open source TLK toolkit for speech recognition features a HMM-DNN (deep neural network hidden Markov model) hybrid architecture for recognition and speaker adaptation using DNNs.

Deep neural networks are a state-of-the-art approach to artificial intelligence and machine learning based on the premise of mimicking the processes of the central nervous system (in particular, the brain). They are currently being developed and applied to all manner of real-life contexts, from data processing to robotics to medical diagnosis to online gaming… And in TLK we use them to identify patterns in speech and language, in order to generate evermore accurate recognition hypotheses and, therefore, evermore accurate subtitles.

TLK is currently available for easy installation in Mac OS X, Ubuntu or Debian operating systems. It can be downloaded here, along with all the documentation and tutorials you need to get started.

Need subtitles for your video repository? Contact our project coordinator for details on our transLectures subtitling service.

Leave a Reply