How To Transcribe A YouTube Video

A few Days ago, someone asked me! "I have some university lectures saved on YouTube. How can I convert these into text?" How to transcribe a youtube video?

I replied: "So, this is a really interesting question. I’ve looked at how to download videos from YouTube before. I’ve dissected all the ways you can convert a video file to an MP3. But we’ve never looked at how to convert a YouTube video to text." It turns out, it’s surprisingly easy, with a couple of caveats. Here’s how to do it in the browser, on your computer, and with the help of someone else.

The Firebug Way


This approach requires you use the Firefox browser. If you haven’t already got it, download it. If you haven’t used it in a while, you should update it. This approach was tested with the latest version of Firefox (40.0), on OS X 10.10.5.

Then, download and install Firebug. For those who haven’t come across this before, this is a tool frequently employed by developers when creating websites. It allows them to dynamically tweak the design, markup, and structure of a webpage, test JavaScript code, and debug any problems. But it’s also got a number of uses outside of web development. Don’t worry if you’re not a coder. You can still follow along with this tutorial.


It’s worth pointing out that there’s a version of FireBug for Chrome, IE, Opera, and Safari. This spin – called FireBug Lite – doesn’t work with this tutorial. You have to use Firefox.


Once it’s successfully installed, open it and click ‘Net’. By default, the Net Panel is automatically turned off. You’ll have to activate it.

transscription-firebug-enable

Then head to the YouTube video you wish to transcribe. Click on CC, and pause the video.

YouTube can also translate captions in real time, although the accuracy isn’t great. If you wish to get a transcription in a foreign language, click the gear icon, then “Subtitles”, select “Translate”, and choose your language.

transcription-firefox-languages

Back in the Net tab, you’re going to need to search for “timed text”.

transcription-timedtext

Once you’ve found it, click it. In the drop-down, select “Request”. This will contain the entirety of your transcription in an XML format.

transcription-request

Select it, and paste it into your favorite text editor. Then get prepared to do some serious tidying up. The YouTube auto-transcriber is questionable at best, and in all of my tests, it produced some pretty strange stuff.

transscription-textpad

You said you’re planning to use it in lectures, however. This might be a less noisy environment and therefore produce better results. As always, your mileage will vary.

Don’t forget, some lectures come with pre-written subtitles. This means you don’t have depend on the ones auto-generated by YouTube. You can use this method to gain access to them.

With Express Scribe Free


I feel the Firebug approach is the best one. It’s free, and despite some dubious transcriptions, it works. Although it’s certainly not the only way.

There are also some free packages that make it easy to transcribe audio files, either by hand or using the built-in speech recognition software in Microsoft Windows. One of the best I’ve come across is Express Scribe Free, available as a free download for OS X and Windows.

This is a professional-quality software package, used by people who actually work as transcribers. If you get frustrated with the quality of YouTube’s automatic captions and want to actually manually transcribe your own lectures, this is for you.

There’s only one prerequisite: you will need to convert your YouTube video to an MP3. Then, you’re ready to start transcribing.  The version for OS X isn’t too dissimilar to the Windows one. It allows you to drop an audio file in, and control it in a way that makes it easy for you to accurately take a record of what’s being said. But there’s one downside: it doesn’t allow you to use OS X’s built-in voice recognition software.


The Windows version does let you use the built-in voice recognition, but don’t expect much from it. It’s still very much mistake prone. For more information, check out Ryan Dube’s detailed run-down of Express Scribe here.

Pay Someone To Do It For You


Of course, there’s also a third option.

Depending on how your budget is, you could get someone to transcribe your document for you. This doesn’t necessarily have to be expensive. On fiverr.com (a popular services marketplace where tasks start at $5), there are 458 different vendors of transcription services.


Some of the most highly rated of these offer 10 minutes of transcription for the bottom rate of one Abraham Lincoln. Although, with Fiverr, you sacrifice expediency for the price. If you pay the rock-bottom price, you can expect to wait as long as two weeks to get your work done. Although you can pay extra to get the task rushed.

Alternatively, there’s also the likes of PeoplePerHour and ELance. Fellow MakeUseOf writer Harry Guinness depends on the latter for the transcription of his interviews:

“It’s easy to find people online through sites like Elance.com who offer transcription services. Simply post your job to the site and they’ll make a pitch saying how much they can do it for.

When you’re posting the job you want to be clear about exactly what you need. Link to the video and ask them to transcribe the first 15 seconds in their reply. I’d pick the cheapest person who’s profile looks good and their transcription is accurate.

I’ve found that I pay about $20 for an hour or two’s transcription. However, I normally need it done on the hurry up. If you can afford to wait, or you have a lot that needs doing, I’d expect you to be able to find a competent person who’ll do it for around $10 an hour.”

It also goes without saying that transcribing work is a great way to earn money online.



Lower Your Expectations


There are a number of ways to transcribe audio recordings. But the question is whether they’re any good.

I found that the auto-captions on YouTube were simply not good enough. It produced far too many mistakes, often rendering the produced text unintelligible. Ryan was similarly unimpressed with Microsoft’s built-in speech recognition software.

If you want an accurate transcription, then you’re going to have to make some pretty steep compromises. Either you hire someone to do it for you, which can be expensive. Or, you do it yourself, which is time-consuming. The choice is yours.

Comments

Popular posts from this blog

The Best Websites on the Internet

Infantile Amnesia: Why Don’t We Remember Things From Our Infant Years?

Eight Billion Reasons Yahoo Didn't Buy Facebook