Juan Martínez Pérez (SWISS TXT)
NERstar is a tool that has been designed to improve the quality of subtitles in terms of accuracy, display mode, and duration. It is a valuable asset for anyone involved in the production and evaluation of audiovisual content, as it can help to streamline the subtitling and evaluation process and make it more efficient.
One of the key features of NERstar is its ability to correct and evaluate errors exclusively on the text level. This means that users don’t have to deal with timecodes or subtitle segmentation, which can be time-consuming and complicated. By focusing on the veracity of the text, users can make edits or evaluate quickly and easily, without having to worry about technical details.
NERstar also includes several automated processes during import, editing, and export, which can help to enforce limits and optimize the duration of subtitles. For example, NERstar can automatically recalculate timecodes and optimize the duration of subtitles to ensure that they are displayed in a stable and consistent manner.
In addition to these features, NERstar also includes several special features, such as the ability to produce sign language videos, to import video files, to add openers, as well as remove tags, and duplicate subs during import.
One of the most innovative features of NERstar is its implementation of AI to shorten text and provide automatic translation. This is a new approach in the field of live subtitling, as it provides a new way to convert subtitle files into files that are segmented by sentences or phrases. By focusing on sentences instead of subtitles, NERstar can correct sentences that extend over one subtitle, evaluate sentences that extend over one subtitle, and produce sign language videos based on
sentences instead of subtitles. This approach is more representative of how we process information, and so does AI.
NERstar is still in the development phase, but there are already planned activities to evaluate and compare its automatic translation quality with other popular translation services such as Google and DeepL. This will help to determine the effectiveness of NERstar. Another planned activity is a workshop to explore possible program formats that NERstar could be used for production. NERstar will also be implemented in one research project to evaluate its sentence or phrase-based approach. This approach is designed to provide more accurate translations by analyzing the context of the spoken words and providing translations based on the entire sentence or phrase, rather than just individual words or groups of words.
In conclusion, NERstar is a powerful tool that can help to improve the efficiency and accuracy of subtitling. Its innovative features, such as the ability to import video files, produce sign language videos, and implement AI, make it a valuable asset for anyone involved in the production and evaluation of audiovisual content. With NERstar, users can focus on the veracity of the text and let the tool take care of the technical details, resulting in faster, more efficient, and more accurate subtitling.
Juan Martínez Pérez has worked with universities, public and private television broadcasters in Europe as an advisor and instructor in speech recognition since 2008. He is working with SWISS TXT and the German company SANDEC on the development of NERstar as evaluation and post-editing tool in connection with different research projects, such as 5GMediaHub, MediaVerse, IICT, and easier. Furthermore, he is a fingerstyle acoustic guitarist, and releases music under his real name and as Elson Complex. On the different digital music platforms, he has currently around 200k monthly listeners.