Translation technology | 09.12.2009

Accessibility drive: Google launch automatic captions for YouTube videos

Language matters have been generating a lot of interest recently, especially in the realms of web-based or automatic tools, and Google has been taking centre stage once more. First there was the launch of Google Translate, which was followed by the announcement that automatic captions (or auto-caps) would be coming to a YouTube screen near you soon.

This is a significant development for millions of people worldwide for it means that the 20 hours of video uploaded to YouTube every minute are potentially accessible to users who are deaf and hearing impaired. But there are also exciting implications for foreign languages in that all videos could eventually be enjoyed by non-native speakers.

Captions have been available in Google Video and YouTube since 2008 and there are now hundreds of thousands of captioned videos available online. But what is different about auto-caps is that automatic speech recognition (ASR) technology can now be used in conjunction with YouTube’s captioning facility to produce machine-generated automatic subtitles.

However, there are certain limitations – only 13 partner channels currently offer this system, the only compatible language is English, and the system is far from accurate. For example, when using the automatic feature, one company found itself to be selling ‘salmon’ instead of ‘SIMs’ for android phones! As with any speech recognition software, it remains to be seen whether regional accents will have a bearing on word outcome. Nevertheless, this is a huge step forward for social inclusion and increased accessibility for deaf or hearing impaired users. Google openly admit the system’s flaws and state that improvements will be seen as the technology grows over time.

In addition to auto-caps, the launch announcement showcased another new feature called auto-timing and also covered the improvements to internet searches brought about by the advent of auto-caps. Auto-timing works when text files are uploaded without the need for a time code file, and text delivery is automatically timed to the video’s audio track. Again, probably not perfect, but something which will facilitate the use of captions and hopefully make them more popular. But it is the improved search facility which has generated interest in the language world because if a video contains written material, i.e. a caption, it can appear in internet search results – and if this text has been translated into different languages, then it will receive more hits. Furthermore, users can go directly to the point at which this ‘snippet’ of text appears, instead of having to watch the video in its entirety.

This is good news for businesses, news channels or any institutions (otherwise known as ‘channel owners’) with online video content. The United Nations and the European Parliament being just two organizations who already post online multi-language videos and who will undoubtedly benefit from this improved search facility.

These developments have been welcomed by YouTube viewers worldwide and there is no disputing its benefit for deaf or hard of hearing users – it has taken 20 years to develop and is also a project of great personal significance for the main software engineer, Ken Harrenstien, as he himself is deaf.

However, machine generated captions are certainly not as accurate as human made captions and Google state that it will be a very long time before this happens. Google Translate also admit that translation quality varies for certain language pairs: Danish translations were said to have raised some eyebrows at an international meeting and when The Translation People tested the Welsh option, it proved to be quite poor. Nevertheless, in the developer’s own words – this is far better than nothing at all, and this opening up of content which was previously inaccessible to many people, is a step towards making information available on the internet accessible to all.

When watching the YouTube video of the launch, the buzz surrounding the announcement was palpable, and understandably so. This is a move which has widened the global net of social inclusion and given real meaning to the word ‘access’.

Accurate text file translation could increase visibility and accessibility of your product on the internet. If you are a channel owner wishing to translate text files or transcribe audio files – The Translation People can help. Our translation services cover over 140 languages and we only employ translators who work exclusively into their mother tongue.

Need help with a translation?
Get in touch with us

Whether you have a specific project you want to discuss, need a translation quote or simply want to discuss your requirements, do not hesitate to get in touch with us.

Get in Touch