YouTube Announces 1 Billion Videos Are Equipped with Automatic Captions

YouTube has just announced that it provides automatic captions on a whopping 1 billion videos. While this is impressive, the real work lies in elevating the quality of these YouTube captions.

History of Captioning on YouTube

In 2006, Google launched YouTube’s first stab at closed captions. While rudimentary, it was a step in the right direction and paved the way for the release of automatic captions in 2009.

Still used today, automatic captions combine Google’s automatic speech recognition (ASR) technology with the YouTube caption system.

Over the years YouTube has worked to improve the accuracy of their automatic captions by 50%. This was accomplished by improving its speech recognition, refining the machine learning algorithms, and expanding the training data. While much progress has been made, YouTube automatic captions still leave much to be desired.  

Importance of YouTube Captions — Not Craptions

YouTube captions provide access to videos for over 300 million people who are deaf or hard of hearing. Additionally, captions allow anyone to watch videos when the environment is noisy, the audio quality is poor, or there are accented speakers. Furthermore, some people would rather read than watch video and others are looking to practice a foreign language.

With poor quality captions, it can be very confusing to follow what is going on in a video. For example, in the trailer for Manchester By The Sea, there are multiple instances of “craptions”.

The first utterance in the trailer is riddled with errors. Instead of saying “if you could take one guy on an island with you and you knew you’d be safe because he was the best man and he was gonna keep you happy…” the captions read “take one guy do and I on with you and you know you’d be safe because he was the best and there’s going to be…”

Later, one character says “Where are we going, to the orphanage?” and the other replies  “Shut up, get in the car.” This is captioned as “we’re going to the often it set up in stock can…”

This continues throughout the video and during a very emotional moment, Michelle Williams’ character is captioned as saying “My heart was broken nose is broken too,” rather than “My heart was broken. I know yours is broken, too.”

As you can see, it is very confusing and frustrating for anyone trying to understand this video by only reading the captions. But for now, this is the best that YouTube automatic captions can offer.

Options for Accurate YouTube Captions

Using ASR to create accurate captions is not an easy feat with the challenges that videos pose. Accents, dialects, fast or mumbled speech, mispronunciations, and poor audio quality all make it hard for automated technology to produce accurate results.

For content creators and publishers who don’t want their viewers to suffer through bad captions, there are two main options.

The first option is to manually edit the automatic captions on each video before you publish. This is feasible for content creators that aren’t putting out a large amount of volume.

The second option it to enlist a 3rd-party captioning service to create 99% accurate captions that upload straight to your YouTube videos. These captions indicate speakers changes and are compliant with accessibility laws that affect many of the videos available online.

For example, the FCC requires any video or video clip that was shown on TV with captions to be captioned when re-shown on the internet.

Another benefit of professional captions is that they provide search functionality and a big SEO boost when utilized properly. This makes it easier for viewers to find your videos and interact with them.

To learn more about professional video captioning click here.