A YouTube caption fail

I’ve been spending some time on YouTube recently, as we’re hosting a number of videos on the platform for a campaign we’re running to recruit foster carers.

Obviously we wanted to make sure we added captions to the videos, and to save a bit of time I’ve been trying out the caption tools. Firstly I noticed that YouTube automatically ‘machine transcribed’ some of the videos. On inspection, though, the accuracy of these was way off – possibly made worse by the particular accents which the speakers had, to which I doubt YouTube’s voice recognition is very well tuned.

YouTube's auto-captioning is way off here

However, the transcript tool fared much better. This lets you upload a plain text file with the transcript of the video. This is far easier than creating captions, which require timecodes.

YouTube then processes the file in a matter of seconds and applies the text as captions to the video, making a best guess at the timing. With the 7 videos I had uploaded, the results were very good, with split second accuracy in many cases.

A bit of user input makes all the difference

One thing that did sometimes throw the timing a bit was where I had skipped some of the dialogue or shortened it for brevity and ease of reading. Overall, though, it was certainly satisfactory and a massive time saver.

In conclusion, then, I certainly wouldn’t rely on the auto-captioning, but it’s fantastic that YouTube are offering this given that the majority of uploaders will never bother manually adding captions. The transcription tool is an excellent compromise delivering better than average results, as it certainly worthy of praise.

I recently bought a new phone. It’s great, but like most smartphones it’s taking a bit of time to get completely used to the complex interface.

On the second day of playing with it, I was browsing the web and came across a site which had Captcha. It was a Google product, and I had to complete the Captcha field to continue with what I wanted to do. The problem was, the image that it had come up with was terribly distorted, and I couldn’t make out the letters and numbers. I tried several possibilities, failing every time. Normally I would simply refresh the page to get a new image, but I hadn’t yet learned how to do that on my phone! I tried going back then forwards again, but it just loaded the same image. And crucially, there was no link to an accessible alternative.

It was one of those rare occasions where I had hit an absolute brick wall because of an inaccessible web design feature.

What did I do? Well, I could have fired up my laptop and done it on that. I could have closed the browser and logged back on, or tried again later. I did none of those things – the moment was lost, as was my interest in completing the task. But what if it had been something really important?

The experience reminded me of a few conversations I have had recently about web accessibility, and all of these had a similar theme – the unnecessary need for some users to take the long way round to access information.

In terms of web accessibility, the most serious type of problem is when a user can’t access information at all (for example, when no alt description is given to an image containing important content, such as navigational tabs). But there is also the slightly ‘softer’ issue where information is available, but the user has to work to get to it.

Are you sending your users the long way round? (Photo by incurable_hippie)

Someone recently posted to the Accessify Forum, asking whether it was a good idea to create an alternative accessible version of a website. I recommended that this be avoided, as it creates an unnecessary division between ‘types’ of user which is undesirable and potentially discriminatory. Making it necessary to click on a “Text-only” or “Accessible” version simply introduces another workaround which some users will have to use, and goes against the principles of inclusion and integration that Web Accessibility tries to promote. It reminds me of when you see signs for wheelchair access which take you round the back of a building – some people will have to take the long way round, and may feel quite inferior as a result.

Another example was during the recent televised UK election debates. When the second debate televised by Sky News hit the airwaves, there was immediate outrage across Twitter and the web at the fact that no live captions or BSL interpretation were available. This meant some users had to look elsewhere for the content – something they should not have had to do.

THERE ARE NO #SUBTITLES (why not?) on #leadersdebate it is NOT #accessible for 1 in 7 (70,000) #DEAF people @skynews you #fail!!! pls RT

An angry Tweet from Pesky People

Too much of the web involves these sort of workarounds for some users, and the problem seems to be that some developers think that this is good enough. Going back to the building analogy, there are often very reasonable factors which mean ramp access has to be round the back – for example, to comply with building regulations. However, there are far less reasons why a web site should require someone to take the long way round.

So if you’re a developer, give consideration to how you’re presenting your content and think carefully about whether you’re offering an equal experience for all your users. Forcing someone to take an alternative route or employ workarounds to get to your content, just as with my experience with Google on my new phone, may well mean they give up and go elsewhere.

Photo derived from a work by incurable_hippie used under the Creative Commons Attribution-Noncommercial license.

I hope to have a closer look at these features in the near future, especially as I’m currently involved in putting sign language videos onto YouTube. For now, though, have a look at the video below for more info.

PS thanks to Phil Teare for the original heads-up about this over at Accessify Forum.

Screenshot of a BSL video

Some of you will know about my project to get British Sign Language content on our corporate website. For the background to this, see the original business case and my experiences in the studio filming the videos.

Last week I got the BSL videos from our audio/visual unit, with all the audio editing complete, bringing us a step closer to getting them online. All I needed to do was add the captions and run them past our BSL expert one more time, to make sure nothing had been lost in translation.

With impeccable timing, Henny Swan blogged about captioning earlier today so I’d encourage people to read her post too as I’ll be referring to her findings here as well.

Online captioning tools

Due to my employer’s IT policy, which means I can’t install software onto my work PC, I’m going to have to settle for online captioning tools. This limited me somewhat, but I was hopeful that I could find something suitable.

YouTube itself recommends CaptionTube, but I’d also heard good things about YouTubeCC, so I decided to give that a go first.

YouTubeCC

YouTube CC

To start, make sure your video is uploaded and processed on Youtube. Next, go to YouTubeCC and enter the URL of the video that you wish to caption. This will bring up the video on the left with a box for the captions on the right.

The idea is that you play and pause the video, tapping in the captions as you go and adding them to the box on the right. YouTubeCC automatically adds the start times, according to where you’ve paused the video.

The first issue I encountered was that there seemed to be no way to preview the captions on the video itself, as you enter them. This makes it impossible to judge whether you have given each caption enough time on-screen, or whether you need to split them up.

Bigger problems were to come, though, upon completing my first video. Clicking the Export to YouTube button brought up the following error message:

Warning:
Some of the captions/timecodes are bogus and will not be added to the export.
Do you want to export anyhoo?

I have no idea why the captions or timecodes should be ‘bogus’, and could see no obvious error, so I went ahead and exported anyway, which enabled me to save the captions as a .sub file (YouTube requires either sub or srt file formats).

Adding captions using YouTube CC

Finally, heading back to YouTube and clicking on the Captions buttons under the video, on the My Videos page, it was a simple case of uploading the .sub file. The results weren’t bad – no sign of the bogus captions or timecodes – but as per my first concern, the timing was not perfect and some captions were far too brief. Also, captions remained visible until the next one came along, which in some cases was not ideal. I realised that you have to insert blank captions to achieve breaks between captions, which seemed a little convoluted.

Overall, then, I found YouTubeCC not bad but lacking features necessary to get really tight captions. Perhaps I wasn’t using it properly – I’d welcome suggestions or advice!

CaptionTube

Caption Tube beta

Next I thought I’d go back to YouTube’s original suggestion, CaptionTube. Henny has already done a great job of reviewing the pros and cons of this one, but I thought I’d add a bit more about some of the things I liked and disliked.

I really liked the timeline at the bottom. Anyone familiar with editing software such as Final Cut Pro will be very comfortable with this kind of interface. Unfortunately, it appeared to lack some of the functionality you might hope for, such as the ability to drag, drop or resize segments.

Caption Tube's timeline

Another feature I liked was the ability to set the duration of a caption. A fairly essential feature, I’d say, but having seen the difficulty of doing this in YouTubeCC I appreciate its existence here.

A definite down side was the need to swap between two pages to edit the captions then preview them. This slowed the process considerably, and although the Preview option is very welcome, I can’t help thinking it could have been integrated more usefully into the main view.

Caption Tube requires you to go to another page each time you want to preview your captions

A final plus, though, is that Caption Tube connects directly to your Google/YouTube account so there’s no need for a separate login.

Overstream

Overstream

Next up, then, is Henny’s preferred option – Overstream. Unlike the previous two, this one requires that you create an account. There’s no obvious privacy policy which is not great, and the password length can only be 10 characters long, far shorter than my usual hyper-secure passwords! Nevertheless, I duly sign up.

The first problem I encounter is the Overstream doesn’t appear to be able to access videos that are not yet public on YouTube. I guess this makes sense, although it wasn’t a problem for the previous two sites. Even using YouTube’s Limited Access URL option doesn’t seem to work. I don’t really want the video to be public yet, but to use this service I don’t have much choice (and can always hide the video again after adding the captions), so I temporarily make the video public.

Overstream has a handy demo video showing you the main controls so after watching this I plunge right in.

A screenshot of Overstream's interface

It’s immediately obvious that many of the issues I had with the previous two services have been addressed here. You can see the captions immediately appear on the video, each caption is given a default length which you can easily adjust, and you can drag and drop each individual caption on a global timeline.

Once finished, you can save the captions, or ‘overstream’, for editing at a later date, and can then export as a .srt file.

Uploading the .srt to YouTube worked a treat, and the timing was much better than before.

Conclusion

In conclusion, Overstream appears to offer by far the best online tool for captioning that I’ve used so far. The other options are ok, but far more basic and not without their faults.

I can’t show the results just yet as I promised to let our BSL signer have a look first, before it went public. I’ll be sure to update my blog as soon as the completed videos are online.