Participatory Culture Foundation

On April 5, 2012, the Universal Subtitles is now ‘Amara’post on the Amara blog, linked on top of all Amara pages for a while, announced the change of name from Universal Subtitles to Amara. This post evoked the importance of community development and also said:

In the coming weeks, we’ll be making the biggest announcements we’ve ever had

11 weeks on, these biggest announcements haven't materialized. The Amara blog has actually been completely silent and it is only since June 22 that Amara, via Darren Bridenbeck, has started giving feedback to users in this present help forum. Otherwise, Amara's silence on this and the previous help forum between April 19, when Marisa Jean Browne made her last reply on behalf of Amara, and June 22, has been impressive. Not even an official announcement of this change of forum, which resulted in some people continuing to use the old one.

Now in his June 22 answer to Richard Gresswell Can't upload subtitles directly todaypost concerning the removal of the possibility to copypaste a plain transcript, Darren Bridenbeck wrote:

Uploading unsynced txt files should be available in the next release of Amara. Keep an eye out for this to come back in the future.

Moreover, concerning the Amara-imposed moderation of subtitling in videos added to teams that do not foresee any such moderation, in a June 17, 2012 e-mail to me and to the owner of the Music Captioning tea, Darren Bridenbeck had written:

Our developers are working on resolving the root issue to these mysterious tasks, and blocked subs (via the DRAFT / non-clickable "edit Subtitles" link) soon and are slated to have it fixed sometime this coming week.

By the following Sunday - June 24 - that root issue not only had not been fixed, but had evolved new ways of crippling collaborative subtitling via imposed moderation. See, on this forum:  New effect ot the paralyzing Amara task-assigning / draft status slapping glitch, Amara-slapped task for Amara-deleted subs? and  53 lines of subtitles showing as "(0 Lines) (in progress)" in the navbar.

So Darren Bridenbeck wrote me yesterday (June 26), about the Amara-deleted subs:

We're looking to have these issues fixed in our next release of Amara

but as in his reply to Richard Gresswell's post, he did not specify a date for this release.

So when is this release going to take place? Considering the crippling effect on collaborative subtitling of the Amara-imposed moderation, it is important to know when it will go away, because it is also hitting subtitling in Amara video pages added to a team by people who did not publish the original, and cannot therefore have their normal subtitling held hostage much longer by this glitch.

And about testing: when it was still called Universal Subtitles, Amara had some kind of testing by users of future releases: the https://github.com/pculture/unisubs/wiki/Dev-Center page still indicates a "Testing mailing list (for anyone helping us test)". But there has been no activity on that testing mailing list since March 20, 2011.

It is particularly important that  this new release be tested by normal Amara users, not only by geeks who may know the sofware well, but seem to have little idea of what we users actually do with it, if we go by the annoying to deleterious effects of recent changes imposed by Amara, apparently without testing by Amara users:

• Annoying: the already mentioned removal of the possibility to copypaste a plain transcript, to be replaced by the possibility to upload it at the next release: if people have to prepare a .txt file for uploading a plain transcript, they can just as well copypaste its content, which is simpler. And Subtitle-Horse.com allows both ways - copy-pasting and uploading - so why not Amara? and why the more complicated way?
• Annoying: hiding the link for declining a task (and for team admins, also hiding the link for reassigning tasks), especially now that moderation tasks are imposed by the Amara software on subtitles of videos added to teams that did not foresee moderation: this results at times in tasks being imposed to people who do not even participate in the team, but collaborated in the subtitling of a video added to it. I mentioned the hidden link for declining a task in Amara-slapped task for Amara-deleted subs? (but only found out about the hidden link for reassigning tasks afterwards)
• Deleterious: the new 32-character line limit, discussed in  Can't upload subtitles directly today, where Darren Bridenbeck wrote "We changed from 40 characters per line, to 32 because it's a standard for captioning; however, this 32 character break is on the Amara display only, and we never add breaks to the actual srt output." Which is the exact contrary of what users have been requesting: in an online video player, the 32 character "standard" does not obtain, but it does if you are making subs for a DVD. So what users have repeatedly asked for is the possibility to insert linebreaks themselves, where they - and not the Amara software - want, in a way that will stick in the .srt file. Moreover, the new Amara-imposed limit is retroactively amputating words in subs in the player that were previously shown whole.
• Very deleterious: whatever the Amara developers changed in the software part concerning teams, which resulted since June 5 in the imposed moderation on teams that did not want it, cannot work properly with it, and where this moderation in some cases block subtitling of videos added to the team, but not put online by team members.

Moreover, some updates concerning teams in the file for translating the Amara site on Transifex are worrying:

• The possibility to override permissions specified in a team's settings, in its projects' settings, is bound to create an unholy confusion for participants.
• The possibility to not only remove a video from a team, but to delete altogether its Amara page. Team owners should only be allowed to do that if they published the original video and created the Amara page for it themselves.

So if any of the above-listed annoying to deleterious features is meant to be included in the new release of the Amara software, users should be allowed to test it before it gets imposed on them.

 

I now agree with you, Verone, because 2 days ago, Amara seems to have stopped whimsically imposing moderation on subs of videos added to teams that don't want moderation (there are still old tasks that go on crippling collaborative subtitling e.g. in videos that happen to have been added the Music Captioning team, though). But when I wrote the post, that whimsical crippling of collaboration was a very serious problem, also preventing translation of perfectly usable but moderation-blocked subs.

The odd thing is that this moderation issue and the line length and unsynced transcript issues were meant to be solved in the new version of the software, and only the first has been partially solved. And there hasn't been an announcement concerning the new version.

Amara only used the announcement feature to invite users to translate the subs of two videos uploaded, already captioned in English, to YouTube,  by GoogleDevelopers. It's interesting to see how the automatic transfer of these captions to Amara messed them up, because of Amara's 32-character limit combined with its lack of solid implementation of line breaks.

For instance, in "Project Glass: Live Demo At Google I/O", if you compare the subs in the original YouTube video with their transfered appearance in the English subs of the Amara page for that video:

• Sometimes Google Developpers multiline formatting via line breaks is preserved in Amara, but the 32 character limit transforms a 2-line sub into an invasive 4-line one.
  
  
• Sometimes this original formatting is differently messed up in the Amara version: e.g at 1:22, the sub in the YT original reads:
  Hey, J.T., J.T., can you hear me?
  >>> Yes.
  But what appears in the Amara player is:
  Hey, J.T., J.T., can you hear
  me? >>> Yes.
  because Amara imposed a wrapping after "hear" and deleted the line break before >>>.
  
  
• Sometimes the original subtitle gets amputated in the Amara player.
  E.g. at 3:42, the YT original sub reads:
  all really good. They're trained. And I have
  great confidence. But this is -- this is a demo that --
  >>> Yeah.
  But ">>> Yeah." is cut off in the Amara player.
  
  Then at 7:27 the YT original sub reads:
  >>> Have fun.
  >>Sergey Brin: It's pretty high up there if
  you haven't been there before. Don't try this
  But "Don't try this" is cut off in the Amara player
  
  The biggest amputation occurs from 9:56, where the YT original sub reads:
  >>Sergey Brin: All right. Now maybe we should
  -- well, all right. Come on up, guys. You're
  already here. Come on up, take a bow.
  [ Cheers and Applause ]
  and the sub stops at "You're" in the Amara player, cutting off its last 2 lines: see the attached screenshot from the Amara player, with the Amara sub dragged above the original YT one.

 

I'm not quite sure what the goal was of asking users to translate these subs in Amara.

 

The result for this video is a good illustration of the facts

• that YouTube allows correct multiline subtitles, with line breaks introduced by humans where it is logical to have them,
• that when such correct multiline subtitles get transfered together with the video to an Amara page, they get variously messed up by a combination of the Amara arbitrary software-imposed wrapping to produce a mock implementation of the 32-character limit, and of Amara's continued refusal to solidly support logical, human-decided line breaks, which is the correct way to produce multiline subs.

Of course, as Darren Bridenbeck explained in Can't upload subtitles directly today , the original sub layout is preserved in the Amara .srt file (well, as long as no one edits them in Amara). However, it is not preserved in the .srt files of the translations: see the attached .srt files of the Amara English and Italian subs of "Project Glass: Live Demo At Google I/O"

 

Logical conclusion: if you want to preserve the original layout when you translate these subs, don't use the Amara translation widget, but translate the .srt file directly, with Notepad, Piratepad or any other desktop or online text editor (1).

Of course, if you then upload the translated subs to Amara, they will still get messed up in the player by the Amara-imposed 32 character wrapping, like the English subs. But at least, it will be possible to download the correctly formatted .srt file and add it to the original video, should the Google Developers wish to do so.

 

Somehow, though, having to do that to preserve the layout seems in contradiction with Amara's claim to be "The easiest way to caption and translate any video." So once more, can Amara developers pretty please:

• give up their imposed mock implementation of the 32 character / line "standard" by forced line wrap
• offer solid support for multiline subs created by human-decided line breaks, as users have been requesting for months, and as is the correct way to create them?

 

(1) Or if you have a Google account, you can translate the original .srt file with Google's own Translator Toolkit. You'll have to reintroduce the manual line breaks in your translation, but they'll stick in the exported .srt of the translated subs. Besides, the Translator Toolkit also provides an automatic translation: it is unpublishable as it is, but good enough in most cases to be editable in less time than required to do the translation from scratch - contrary to the Microsoft automatic translation offered in the Amara translation widget, which is usually just totally unusable.

Thanks to naomib who told me about the Toolkit when we were working together on the English subtitles for Android 4.0 Accessibility Demo: Turning on Accessibility: it also proved invaluable for translating original subs that though complete, were made unusable in the Amara translation widget by the Amara-imposed moderation.

Correction: Amara didn't stop slapping arbitrary moderation on subtitles after all: it just did on my revision of the French subtitles of President Obama's speech at NALEO. Banner saying (for me):

Draft only Subtitles translated from English. Showing Revision 6, created Today by Claude Almansi .

with a task assigned to me erroneously saying :

Video: Watch President Obama's Full Speech... (27:40)

Subtitles: latest draft translated by Laura Emack

with the link on "latest draft" leading to Laura's previous revision, which is also the version shown to others than me when opening the French subs.

In this obsolete version "for others", the Edit Subtitles works, but when you click it, there's a message saying:

These subtitles are moderated. See the NewsHour: Vote 2012 team page for information on how to contribute.

which at least prevents others from editing the obsolete version, but nevertheless this new avatar of the Amara-imposed moderation mess belies again Amara's claim to be "The easiest way to caption and translate any video."

 

Hi Verone,

The messed up accented letters in your example might be due to the character encoding of your text file (I mentioned that possible issue in my reply to Darren in Can't upload subtitles directly today).

I.e: when we copypasted transcripts, Amara automatically encoded them in UTF-8, which shows accented letters correctly. Now that we must upload transcripts as .txt files, we must make sure we encode them in UTF-8 ourselves.

As a counter-example to yours, I downloaded as .txt the subs I'm presently working on, choosing them because they have particularly high number of both French and Greek accented letters. Of course, being downloaded from Amara, the .txt file was automatically in UTF-8. Then I reuploaded the .txt file as French in  universalsubtitles.org/videos/xHZzawTR9MLF/info/long-blank-video-14639 (1), and the corresponding "French" subs show all accents correctly.

Darren, what about an explanation about this UTF-8 encoding in the upload widget? Or maybe linked to in the upload widget if there is not enough room for the full explanation?

Best,

Claude

(1) That long-blank-video-14639 page is an Amara page I made for trying things with translations, etc. As the name says, the streamed video is completely blank.  Or at least was until I slapped annotations on it to promote videos I like. Anyway, everyone is welcome to use it as a kind of sandbox too.

Other problem with the upload txt feature: I did that with the English transcript for the Large Hadron Rap video page, and Amara made 2 sets of English subs. Could one be deleted, please.

Nice thing: Amara didn't spawn a moderation task for the resulting subs though I temporarily added the video to the Music Captioning team before adding the transcript.

That gave me an idea for a workaround to avoid moderation: I added another video to the team -  Could supersymmetry explain everything? - for which I had no transcript. But I uploaded the attached 3-line .txt file as if it were one, ticking the box for "This subtitle file is 100% complete", and again, Amara didn't slap a moderation task  on the resulting English subs (now Revision 0 - Today).

I then deleted the subs of that Revision 0 and made 3 real ones, then did Save and Exit: again, no Amara imposed moderation.

Last time I suggested a workaround to avoid uncalled for moderation (Don't want Amara-imposed "Draft status" blocks? Lie to the software - i.e stating that subs made in the widget covered the entire video even when they didn't), it only worked for a day, then the Amara software started slapping moderation on the subs marked complete, then preventing marking complete subtitle sets even if only the last sub was not end-stopped etc.

Let's see what happens with this one: moderation on subs in teams that don't want it and haven't configured it in their settings is a glitch developers are working to remove, isn't it? So they shouldn't object to telling the glitch white lies to make it pass its way, normally.

I see your point, Verone. In an off-forum exchange with 2 Amara people, I tried to convey to them that describing a software issues in tech term can be done simply and concisely, but that the impact of such issues on humans striving to collaborate is much more complicated.

I hope they roll back the copy-paste possibility. Unless - but that seems unlikely - they can make the importing part of the software omnivorous, i.e. capable of properly digesting other languages than English with other encodings than UTF-8.

PS I forgot: there is another argument in favor of copy-pasting: OK, IOS devices can't use Amara at all, but Android devices can. And with an Android device, you can copypaste a transcript, but you can't produce a properly UTF-8 formated .txt file