Automatic captions was first launched by Google way back in 2006. In 2009, Google made an announcement stating that it will be introducing the automatic captioning system to YouTube, to help those who are deaf and mute enjoy the videos better.
Though initially, the results were not as expected, but with the passage of time Google’s coding system got advanced and churned out great results. Although the system is still not perfect, no one’s complaining.
The video sharing service on Feb. 16 stated in its official blog that currently around 1 billion videos have been given automatic captions by YouTube. These 1 billion videos are watched 15 million times per day by users around the world.
According to Google, with all the improvements so far, it has been able to get relatively closer to “human transcription error rates”, and in turn the enhancements made so far have showed a 50 percent increase in terms of accuracy in English. A link to an older video released by Google in 2009, shows how the captioning system worked during the time.
How Were The Captions Added?
“One of the ways that we were able to scale the availability of captions was by combining Google’s automatic speech recognition (ASR) technology with the YouTube caption system to offer automatic captions for videos,” wrote Liat Kaver, the Product Manager of YouTube.
Initially there were some limitations and the results churned out were not per the expectations of the team. This prompted some creators to have a little fun at YouTube’s expense.
However, the team at YouTube was successful in achieving its main goal, which was accuracy for automatic options. Given that diverse content is uploaded on YouTube every day, it was quite a challenging task for the coders to gain maximum accuracy.
Kaver states that YouTube achieved this monumental task by improving speech recognition, expanding training data and also improving the machine learning algorithms.
The Next Step
The goal of YouTube still remains the same – achieving absolute accuracy and going beyond the 1 billion mark. The automatic captions are currently available only in English, but Kaver states that the team is looking forward to extending it to all the 10 languages supported by YouTube.
YouTube is looking forward to collaborating with various creators and viewers all over the world to achieve the level of accuracy in all additional languages. Source: techtimes