Netflix and Large Scale Subtitle Translation

p.p1 {margin: 0.0px 0.0px 10.0px 0.0px; font: 11.0px ‘Trebuchet MS’; color: #000000; -webkit-text-stroke: #000000}
li.li1 {margin: 0.0px 0.0px 10.0px 0.0px; font: 11.0px ‘Trebuchet MS’; color: #000000; -webkit-text-stroke: #000000}
span.s1 {font-kerning: none}
span.s2 {font: 11.0px Symbol}
span.s3 {font: 7.3px ‘Trebuchet MS’; font-kerning: none}
ul.ul1 {list-style-type: disc}

Netflix is working on an incredible, but complex, solution to a very difficult problem. As one of the largest digital media and entertainment companies in the world, Netflix is working on being able to bring its content to users around the world. 

Netflix’s international business will soon account for more than half of the company’s revenue, so the company has begun to invest massively in its ability to translate and subtitle content for users outside the US. However, Netflix is in a difficult position for translation, since all of the content produced by the company (TV shows, and movies) contains the most difficult parts of language to translate, jokes, idioms, sarcasm, cultural references, slang, etc. 

How does one go about handling, and ensuring quality translation on such a large scale for such difficult content?

Last year Netflix revealed the HERMES tool, which allowed thousands of translators in hundreds of languages the ability to test their translation skill, in order to be an approved resource for the company. The test consisted of several rounds of multiple choice questions which were meant to measure the translator’s ability to:

  • Understand English
  • Translate idioms, etc. into their target language
  • Identify linguistic and technical errors
  • Subtitle proficiently

p.p1 {margin: 0.0px 0.0px 10.0px 0.0px; font: 11.0px ‘Trebuchet MS’; color: #000000; -webkit-text-stroke: #000000}
span.s1 {font-kerning: none}

Through the completion of the test (which consisted of thousands of randomized questions, so that no two tests were the same) Netflix was able to produce scores regarding a candidate’s skill level in translation, and match candidates with projects that would suit their skills.
This will also allow Netflix the ability to self-manage and work with translators directly rather than having to go through 3rd parties like us. For an enormous company with deep pockets, this was a rather incredible solution to a difficult problem. Within the coming years we should see Netflix content available in just about any language, as their international sales will no doubt soon surpass their US market.

Leave a Reply