Producing video subtitles is a laborious process. First you must transcribe the original video, writing down everything that's said, proofread and correct then, synchronize the subtitles with the audio so they appear on-screen right when the lines are being delivered. Finally, you translate the text into other languages. Amara is a platform that tries to crowd-source all of this work, making it possible for you to set up a system where droves of volunteers help you produce video subtitles for free, without having to download or install anything. It's not entirely successful, but it's an interesting first step in the right direction.
Before you can translate a video, you must first transcribe it. You can select any video for transcription -- you don't have to own the content: it just has to be available online. Simply provide Amara with a video's URL on YouTube, Vimeo, or another online video service, and it launches into the transcription interface. You don't have to open an account before you begin--you can just start working.The first step in the transcription process is just writing down what the people in the video say without worrying too much about typos and capitalization. Amara's transcription interface is simple and intuitive. By default, it plays four seconds of video, then automatically pauses. You then type what you've just heard, and hit Tab to play four more seconds. If you miss anything, you can hit Shift-Tab to rewind four seconds and listen again. If you don't like to constantly hit Tab and Shift-Tab, Amara can also auto-pause the video for you. In this mode, you simply listen to the video and type as you listen, with Amara pausing it automatically to let you catch up. The way this works isn't clearly explained (Amara calls it "magical"), but it works remarkably well: The video paused and played right when I needed it to, and I had to hit Shift-Tab to rewind only rarely. Even with the excellent auto-pausing engine, transcription is still a laborious process, though. I touch-type quickly, but transcribing a four-minute video took me about twenty minutes of intense concentration.
The next step after transcribing the video is timing the subtitles. Here, you must watch the video and click a button whenever the speaker starts on the next subtitle. Just like transcription, this requires unbroken concentration; fast reflexes help, too. Amara's video introduction to this step says it's "like a computer game," but it's not one I would play for fun. The interface is effective, but it definitely feels like work.
The directions also say you don't have to worry if you get the timing slightly wrong, as you'll be able to correct it later. Accordingly, I didn't worry much -- but when I got to the final step, reviewing and correcting the subtitles, I discovered things aren't so simple to correct. I wasn't always able to extend or contract the subtitles along the timelines so they synced correctly, and the whole process quickly got out of hand. The end result I got reflects Amara's strengths and weaknesses: The video was fully transcribed, but the synchronization was only so-so. Another issue was that some of the subtitles were too long: Amara doesn't offer an easy way to shift text from the end of one subtitle to the beginning of the next (except for manually copy-pasting), so if you happen to break things down into too large chunks when transcribing, you'll have a problem later on.
Sign up for CIO Asia eNewsletters.