Google today announced a big YouTube milestone: 1 billion videos with automatic captions. And these videos are popular: YouTube users watch videos with automatic captions more than 15 million times per day.
Captions are critical for the more than 300 million people who are deaf or hard of hearing, but they’re also useful for anyone looking to read instead of listen. Whether the audio quality is poor, it’s loud wherever you are, the person is mumbling, you want to practice a different language, or you simply want to hear and read simultaneously, automatic captions are a game-changer.
Google first launched YouTube’s video captions in September 2006, then took the feature to a whole new level with automatic captions in November 2009. It therefore took just over seven years to pass the 1 billion mark. That said, automatic captions are far from perfect and a constant work in progress thanks to mispronunciations, accents, dialects, and background noise.
Over the years, Google has been working to improve the accuracy of automatic captions, which is not an easy feat given YouTube’s size and diversity of content. By improving its speech recognition, refining the machine learning algorithms, and expanding the training data, the company has managed to boost accuracy for automatic captions in English by 50 percent. In fact, the company today declared that automatic captioning in English is getting “closer and closer to human transcription error rates.”
YouTube is looking to extend that to all its 10 supported languages. In addition to English, that means Dutch, French, German, Italian, Japanese, Korean, Portuguese, Russian, and Spanish. This is only possible if the site’s creators and viewers participate.
One day, every YouTube video could have an automatic caption track generated and reviewed by the creator. The technology isn’t quite there yet, so until then, you can do your part by reviewing, editing, or unpublishing automatic captions.