transLectures-UPV Toolkit v. 1.3.1

The latest version of the TLK toolkit for automatic speech recognition is available for download from the Tools section.

With transLectures drawing to a close, this will be the last version of TLK released as part of the project, though updates will continue to be made into the future.

TLK, the transLectures-UPV toolkit, is the automatic speech recognition (ASR) software developed at Universitat Politècnica de València. It is made up of a set of command-line tools for building, training and applying acoustic models that can be used to generate transcripts for video lectures. The software behind transLectures-TOT, our “try before you buy” online service, it is also in use at the UPV’s poliMedia video lecture repository. See an example here.

The latest release includes further improvements to the tL-transcribe tool, added in version 1.2.0, as well as a series of bug fixes. tL-transcribe can be used with the ready-to-use transcription system also included in TLK for Spanish media files. The language data made available for this transcription system is limited, and so the results are not as impressive as when the larger, fully-adapted transLectures system is applied. Even so, we think it has its place within the open source community to show how these kind of systems work. Details of how to download and install this system can be found in the corresponding tutorial.

Leave a Reply