Improving conditions for subtitlers and increasing Netflix’s subtitled content

Professor Jorge Díaz-Cintas’s research led to the first subtitling test Hermes, used by Netflix and the media localization industry, to assess the linguistic and translational abilities of subtitlers.

The arrival of streaming video-on-demand services, such as Netflix, Disney+ and many more, has transformed the way in which audiovisual media reaches diverse global audiences, and has radically increased the need for subtitling. This rapid rise in the need for subtitlers led Netflix executives to approach Professor Jorge Díaz-Cintas in 2013 to lead the design, testing and implementation of Hermes, the first online subtitling test and indexing system.

Díaz-Cintas’s research had shown that validation of translation competence should consider not only the candidates’ linguistic knowledge of the foreign and mother tongues, but also the socio-cultural environment in which language is used as well as the technical constraints that characterise the practice of subtitling. As the Chief Coordinator of the 75+ researchers from 40 countries involved in the Hermes project, Díaz-Cintas developed a pioneering approach to testing both established translators and newcomers to the subtitling industry and proposed a set of guidelines to guarantee quality subtitles. The system was designed to assess the skills of subtitlers around the world, covering 32 languages from Arabic to Vietnamese, to ensure high quality standards. The Hermes test provides a good indication of a candidate’s skill level, enabling Netflix to match high-quality translators with individual projects. A year after it launched around 500,000 had taken the test.

Aware that the company was approaching a point where English was expected to cease being the primary viewing experience on Netflix, Hermes was conceived as a pivotal development that would “allow us to better vet the individuals doing this very important work so members can enjoy their favourite TV shows and movies in their language”. As underlined by Netflix, Hermes allowed them “to confidently provide the subtitling of audiovisual content across the 32 different languages covered by the test, upping their coverage from 16 to 32 languages in a single year”. As a result, they were “able to widen [their] global reach and attract new audiences […] Since the Hermes test was launched in March 2017, the Netflix subscriber base has grown from 92 million to 195 million paid members (as of September 2020)”.

Its success has led to an increase in freelance rates. The Hermes system gives subtitlers a competency score (the H number) based on their performance. High H numbers have come to represent a particular standard of service, and this has enabled translation companies and content providers to find and employ the best subtitling professionals.

Working with universities in Algeria, China, Colombia, Estonia, Germany, India, Italy, Lithuania, Peru, Qatar, Spain and Russia, among others, Diaz-Cintas has helped to develop new training courses in subtitling to meet the quality criteria set by Netflix and raise the academic visibility of subtitling in countries and contexts where previously subtitling did not exist as an educational topic. The research has also led to professionalization of subtitling, with improved pay and accredited training courses for subtitlers.

He has given over 120 invited talks on the topic, reaching in excess of 10,000 people, and has led numerous training the trainers events. He has also validated undergraduate and postgraduate courses on subtitling around the world.

Improving conditions for subtitlers and increasing Netflix’s subtitled content

Learn more

Follow @UCLResearch

Follow @ArtsHumsUCL

REF 2021 News