Cheating Subtitles: How to Add Them Easily (and Qualify for Amazon Video Direct)

[@robertBrownboater]

I have a tv series on Amazon called "I Am Second Voyage" uploaded with English CC. I now have SRT files for German that I want to add to it so it would be available in Europe. I can't seem to find a way to upload them to the existing series. Thanks Bob

[@focuspulling]

This is an ongoing complaint against Prime Video Direct (PVD) and their inaction bespeaks a gigantic insensitivity in that department (which moves glacially slow, too). It would take almost zero software platform development effort to allow for multiple subtitle languages per title, but some mid-level employee at PVD probably dumbed things down -- to avoid responsibility -- by siloing English subtitles to U.S. uploads, German subtitles to German uploads, etc. -- even though accessibility itself is about more than just accommodating hearing-impaired persons in each respective "native" nationality (as if such a thing exists in a globalist society). It's shameful, and they know about it, and they clearly don't plan to listen to us. PVD is a "let 'em eat cake" patsy for Amazon to claim that they support indie filmmakers, but it pays out to almost everyone at about a penny per hour (for Prime streaming), and we all know it's a joke.

[@BigDSProductions]

Great to see some of my developing theories in actual practice! I wish I had found this video a few weeks ago as it would have saved me a lot of work (done in my free time gratis). I do part-time work for several community television stations in the area that handle the filming of local town meetings. The new ADA guidelines that go into effect soon have left many school systems and community tv stations scratching their heads wondering how they will implement captioning for municipal and education videos. While professional broadcasters have just thrown money at these requirements for decades, most of the stations I deal with have total annual budgets of less than $100k. And the local school districts are already dealing with huge cuts (here in financially disastrous CT). Your workflow you've presented here is pretty much the same one I have developed as a low-cost solution (obviously there's still payroll hours involved with this workflow). The only downfall is that most of the stations don't have the resources to dedicate an employee to watching complete meetings and correcting errors. So if higher accuracy ever becomes paramount, there may be another option that's a more automated solution using Dragon than your solution of reciting the meetings as you listen to them in headphones. There's an A/V tech guy I spoke with at one school district that has been using Dragon, real-time, with BOE meetings for several months. He said he has had a good deal of success playing the audio directly to Dragon. But it took some work early on to get Dragon's AI up to speed. Once Dragon learned more, it worked pretty good. Of course, he says this was only possible because the meetings usually just feature the same few people every month. I can't speak to this workflow, and I didn't think Dragon would work well with several different speakers, but I'll probably give it a try if the accuracy becomes an issue. Great video! P.S. Please get a quieter mouse.

[@focuspulling]

Thanks for your comments. I recently revisited this issue to see where is the state-of-the-art when it comes to voice recognition. What egged me on was the arrival of two new Premiere Plug-Ins that exploit both the Watson and Speechmatics APIs, in comparison to the Google API that's put to use in the KZbin workflow that I describe in this video. If you Google around for comparisons, what's shocking is that Google still beats all the rest for accuracy -- after all, they're working hard to get those little Google Home boxes into everyone's homes, with good speech recognition, to put a dent into Amazon Echo's dominance. I'm constantly hoping for better results, but my updated assessment is that this free workflow remains ironically the best (and I have no use for deep integration with Premiere in terms of keyword searches, etc.). Regarding the sound of the mouse, it's actually an errant setting in the Camtasia screen-capture software I used, annoyingly adding a digital recording of a mouse click every time I used it! The setting is now permanently toggled off, going forward...

[@MarcoInfante]

How do you add multiple subtitles on Amazon video direct? I don’t want just English but Spanish and French, etc... Can you upload more then 1 .rst file ?

[@focuspulling]

That's a huge and unfortunate limitation at Amazon Prime Video Direct; something complained about for years, and never resolved. It would have taken almost zero effort for them to build that into their interface (it's already built into the Prime Video interface for titles outside PVD), meantime they require completely separate entries for the U.K., Germany, and Japan platforms where they allow PVD. Simply put, it's a mess. One language per entry. Please let them know you dislike this, and there's a 0.01% chance they'll do anything about it (after about 1/2 decade of no action).

[@MarcoInfante]

FocusPulling (.com) thank you! That's crazy that they don't fix that. What do I do if I want to up load a version with Spanish subtitles in South America... what should I do ?

[@focuspulling]

@@MarcoInfante If streaming is allowed from Amazon.com (U.S.) to South America, unfortunately at this time, you can only upload English subtitles (which are necessary to qualify at the U.S. site). There is no PVD on any South America-specific Amazon site.

[@ThomasGrillo]

Thanks for this video. You mentioned a method of using Dragon Naturally Speaking by listening, then talking into a microphone. Wouldn't it be easier to run the audio program from a playback device, directly into the Dragon equipped computer? That way, you don't have to listen, and talk, and go back to do spot corrections so often.

[@focuspulling]

That would be the equivalent of what KZbin already does: analyzing the audio directly from the source. Whereas, the Dragon method leverages its voice learning capability: you have trained the software to understand your voice much better.

[@bluedeath996]

The subtitles for this video on youtube are incorrect in places.

[@focuspulling]

That's so funny; and a great testament to how finicky KZbin's automatic subtitling function can be. At the time when I originally posted this, the automatic subtitles did not generate, but clearly at some point shortly thereafter, they showed up. What I hope for in the future is more transparency from the KZbin backbone about when automatic subtitles appear, when/why they don't, and a manual triggering mechanism too.

[@bluedeath996]

I am glad you also find it funny, I didn't want to offend. I just found it ironic or poetic justice or something.

[@craiginzana]

Any suggestions for alternatives for Mac users? We have a film that I made closed captions for in Adobe but I keep getting this publishing error that says "The captions file contains a TTML error. Please submit a new file." and I'm not sure what that means much less how to fix it :/

[@focuspulling]

Sadly Adobe Premiere Pro even with its newest improvements is widely regarded as having nearly useless captioning functionality. (Their forthcoming, as-yet unreleased version promises minor tweaks, but nothing game-changing.) It is the case that most of the good, free subtitle editors are for Windows - en.wikipedia.org/wiki/Comparison_of_subtitle_editors - but you might want to check out Aegisub from that last for OS X: www.aegisub.org. Lastly, you may want to explore more subtitle file export options if you're using the newest version of Premiere CC.

[@craiginzana]

Thanks for the feedback. It's frustrating that the Premiere software doesn't function properly. I can only image how many other filmmakers have spent so much time only to find out it won't work. I'm thinking about just going with a paid service that's relatively affordable through Rev.com Thanks again for the response! Hopefully Adobe will make more of an effort with this sometime in the future.