You are not logged in.

  • "ossi11111" is male

Posts: 1,066

Date of registration: Nov 27th 2013

Language Team: German

Focus Group: Translator
Translation Proofreader
Final Reviewer
Language Coordinator
LTI Development Group

Thanks: 28643 / 363

  • Send private message

61

Friday, February 7th 2014, 9:37am

@nomada:
So you want to completely replace Dotsub with a more integrated solution. Sounds like a good idea on paper, but it is hard for all the old people to go with that because it's a very drastic change. How do you want to deal with that? This will probably create more chaos for a long time. In Computer Science when a new program or process is developed and implemented at the location of the corporation, it is very hard to show the employees how they have to change their behavior. Most of the time they are not willing to do so.

Concrete example:
Imagine a system for storing physical health records of a hospital. In Germany there is a law that you have to store them several years. So there are many health records in many storage racks. Now a computer scientist is asked how to improve the system, because there is not much space for storage left and it costs much money to store them. Also it takes some time to find the places where the particular health record is stored when looking for it. The computer scientist recommends storing the new health records electronically. But the doctors are so used to printing them out, that they still do it - again and again. The doctors also have not very much time to learn a new system, even if it is better in the long run. They are not willing to learn it. So the biggest problem is not to create the appropriate program but to change the thinking of the users - in this case the doctors. That's the main task!

The others who haven't yet, would you like to share how you wish to help solving that issue of low quality transcriptions done by volunteers who aren't much around communicating and so giving rise to the need for the longest stage in the English transcriptions - timeshifting and restructuring of subtitle strings in the transcriptions where the 'new' volunteers helped?

The way I am addressing similar problems in the German team is to talk directly to the new volunteers and show them how we work, so we all use the same procedures from the start. For new people it is not hard to use a process different from the processes the old members use. For the new people it is new no matter what, so they accept it very easily. Also I am directing them to use this forum as main working location, so everything starts and ends by going to the forum and everyone else is able to read it. It does not work, if somebody uses Facebook, others only talking via Skype, others using the forum. Our main focus is the forum and all relevant information about specific subtitles is written in the relating thread. There you find the pads, the link to Dotsub and the information if the subtitles are completely finished and ready for official upload.

About ossi suggesting that we use pads to proofread, I don't see us making all this more efficient and easy with adding one more platform to work on the same projects that are also worked in other platforms when on different stages of the same language. I think the same when considering across languages (Ex: 1 platform for english transcriptions and another for german translations)

The pads make the communication easier, because they encourage collaboration. To not create an overhead by using pads I was proposing my web tool.

I am realizing now, that every language team uses different processes. I was assuming that you work very similar in the English team as we do in the German team. That assumption seems very wrong now. ^^
Do you work collaboratively on the transcriptions? How do you do it, if it is done offline or in programs which do not support that? How do you decide which version of the transcription is the most appropriate one? How do you maintain the old corrections for comparing it with the new ones? That is why we use pads in the German team.

How are all those volunteers in subtitling and translation forums doing their subs, many times with good synch and string structure? And how come they make so many?

In the German team we use the transcriptions of the English team and build upon them.
Please tell me how outdated this video is: YouTube
That's exactly how I thought it works before reading this thread. :D
Signature from »ossi11111« Willst du beim Übersetzen mithelfen?
E-Mail an: GermanLingTeam@gmail.com

Bisher veröffentlichte Videos mit deutschen Untertiteln:
GermanLingTeam - YouTube

1 registered user and 28 guests thanked already.

Users who thanked for this post:

Ray

  • "Christos Pap." is male

Posts: 721

Date of registration: Nov 21st 2012

Language Team: Greek

Focus Group: Final Reviewer
Language Coordinator

Location: Thessaloniki, Greece

Thanks: 24082 / 76

  • Send private message

62

Friday, February 7th 2014, 8:59pm

Ossi, your concrete examples help me a lot in understanding your points! Keep it up!

Rod, can you restate that part about Amara? What exactly is your vision? Are you thinking of letting other language teams use Amara? If so, how will they use both Dotsub and Amara? You do not want Amara for the LTI because it has no restrictions?

I am currently experimenting on both VSS and Amara because they seem to be the most popular. I'll give you feedback when ready.
Signature from »Christos Pap.« To come to the meetings, install teamspeak from HERE and then press HERE to be transferred directly, and look for the "Linguistic Team" Channel.

35 guests thanked already.
  • "Christos Pap." is male

Posts: 721

Date of registration: Nov 21st 2012

Language Team: Greek

Focus Group: Final Reviewer
Language Coordinator

Location: Thessaloniki, Greece

Thanks: 24082 / 76

  • Send private message

63

Friday, March 7th 2014, 8:32am

Guys, you might want to check out Rodrigo's feedback about Amara testing. There are a lot of possibilities on this tool and a lot of things that can make our lives easier such as integrating timeshifting in transcribing.

Testing Amara for proofreading/transcribing
Signature from »Christos Pap.« To come to the meetings, install teamspeak from HERE and then press HERE to be transferred directly, and look for the "Linguistic Team" Channel.

1 registered user and 31 guests thanked already.

Users who thanked for this post:

nomada

  • "Christos Pap." is male

Posts: 721

Date of registration: Nov 21st 2012

Language Team: Greek

Focus Group: Final Reviewer
Language Coordinator

Location: Thessaloniki, Greece

Thanks: 24082 / 76

  • Send private message

64

Wednesday, March 19th 2014, 11:51am

Case Study: Aegisub

(This is a beginner's feedback but not everyone here is an experienced timeshifter/transcriber, so this feedback might help people in my situation.)

For those not in the know, Aegisub is an offline tool used by some people in the English team in order to timeshift subs. It is an open source software which is already installed in your Linux if you use Mint.

Now, I noticed that the video mode is not functional (at least on my pc). The video's flow is very bad, so you can't use the video mode in order to timeshift. I am guessing that experienced users use the audio mode (which looks similar to VSS) but for a newbie the video can be very helpful since you can have a visual sense of how much gap there is between strings. And you need this if you are a newbie and can't figure out the necessary gap between strings just by numbers or from listening.

Now, the audio mode seemed a little hard to me as a beginner. I couldn't find a way to set the start and end of a string other than pause the audio, check the start and end times of the selected part and then copy these times on the string's start and end times. Is there any other way to do it like in VSS where you drag the starting and ending line wherever you want. If not, then the only way to set the start and end times of a string is to write the related numbers on the start and end frame of each string.

There were features which seemed helpful to me, such as merging or splitting strings, spell checker...About spellchecker, I noticed that you have to do it for each string separately. Is there any way to have it enabled all the time? There is a feature that help you find the string you are looking for at once, and you can see the duration of each string which is important to timeshifters. However, I did not see info about the gap between strings, an issue which is also very important for timeshifters, and I know that you have certain guidelines about it. VSS actually has this feature along with how many characters per second are presented on the screen, which is also a great tool to get you to feel how a viewer sees the subtitles (eg. too fast, too slow, good flow). Also, I am not sure whether you can preset your own "rules" in timeshifting like you can do in VSS, where you can install from the beginning the guidelines of the English team for timeshifting. Therefore, I am not sure if it can help us make the transcription/timeshifting stage faster or even obsolete.
Signature from »Christos Pap.« To come to the meetings, install teamspeak from HERE and then press HERE to be transferred directly, and look for the "Linguistic Team" Channel.

2 registered users and 30 guests thanked already.

Users who thanked for this post:

nomada, Borislav/lizardman

  • "Christos Pap." is male

Posts: 721

Date of registration: Nov 21st 2012

Language Team: Greek

Focus Group: Final Reviewer
Language Coordinator

Location: Thessaloniki, Greece

Thanks: 24082 / 76

  • Send private message

65

Wednesday, March 19th 2014, 12:01pm

Case Study: Amara

Amara (formerly known as Universal Subtitles) is an online open source translation platform for videos. I played with it a little, and while it seems a lot of faster and easier than Aegisub and VSS and it can help us integrate the different stages of English translation, I will not expand on it at this point because it is an online platform (and a pretty open one) and there is an important issue to consider before using it. I can't suggest using it without dealing with this issue first. Ok, here goes.

I experimented a little more with Amara and I just found out that the option that Rod talked about, the one where you can't actually edit a translation but just create a new version of it, applies only on the beta version. For those of you who don't know, in Amara you can currently try the new beta version. In the beta version, every subtitles you choose to edit (if they have been already created), you don't actually edit the same subs. You just use these subs as a reference (or a basis if you will) and you create a new version of them. Therefore, you cannot edit or delete an already finished subs version. You can only create new subs versions unless of course you want to create a new version of a new language. However, this applies only on the beta version, yet. This means that on the current version of amara, there is no way to guarantee that no-one will change your translation or that he/she will keep working on after you have moved it to Dotsub for proofreading, like we do now. This might a create a duplication of efforts if we use it to transcribe/timeshift videos since our trascribed/timeshifted subs will stay there (and in Dotsub too). I personally used it to create some timeshifting in order to compare it with the other tools used by English timeshifters (eg. VSS) but now I am sensing that we may create complexity and duplication if we use it like this. Any ideas on how to deal with it?
Signature from »Christos Pap.« To come to the meetings, install teamspeak from HERE and then press HERE to be transferred directly, and look for the "Linguistic Team" Channel.

2 registered users and 27 guests thanked already.

Users who thanked for this post:

nomada, Ray

  • "Christos Pap." is male

Posts: 721

Date of registration: Nov 21st 2012

Language Team: Greek

Focus Group: Final Reviewer
Language Coordinator

Location: Thessaloniki, Greece

Thanks: 24082 / 76

  • Send private message

66

Sunday, March 23rd 2014, 4:50pm

Case Study: VSS

Visual Sub Synch is an offline tool which is usually used for time-shifting videos. It can also be used for transcription and proofreading but I will focus on time-shifting for now. You can find a video tutorial for VSS here or if you prefer the text version, I have already uploaded it in this post. Having already tried amara, aegisub and subtitles workshop, I thought I'd play a little with VSS in order to see if there any advantages which can help make our efforts more efficient. So, here goes the feedback.

First of all, a very useful feature of VSS is that it allows us to establish our time-shifting preferences (like how much space is allowed between 2 strings) right from the start so that we won't have to worry about applying them manually. So, before starting we can set it up so that it applies our preferred rules in every string (eg. how many characters we would like to see per second or what are the minimum characters per second). Now, the video mode was much better than Aegisub. In Aegisub, the video flow was terrible (at least on my pc), which means I couldn't see in the video the strings that I had created. I am insisting on this because as a beginner, it is very useful to also see the video in order to have a good sense of string length and duration. I guess that if you are an experienced time-shifter, the audio is enough for you. As far as timestamps are concerned, in VSS you adjust them by dragging and dropping certain lines which appear on the top of your screen. These lines constitute the time frame of each string, so you can adjust the string's start and ending time by moving these lines. Plus, you can zoom in and out on these time frames, which means that you can make the string beginning and ending pretty acurate. Of course, for deadly accuracy, you can always manually type in the start and end times on another time frame which VSS provides, where you can actually see the exact numbers. However, I had a problem. In the written VSS tutorial it reads "On
the right side of the strings, in the RS column, VisualSub gives us
information on how many characters/second each string has
...". I couldn't actually find this RS column, so any help could be great. Also, there is a "check errors" box which checks all the strings at once for possible errors and instantly it gives you a list. Last but not least, VSS gives us the option to save the subs in UTF8 format unlike Subtitles Workshop. If the subs are not in UTF8 format, some characters cannot be read by Dotsub. Oh, and there is also a wonderful correction tool which you can find on the tutorial I uploaded, which can check your subs when you are finished time-shifting for any strings which do no comply to the time-shifting rules you have set while editing your preferences.

Now, is anyone willing to give me an online tutorial on VSS and maybe Aegisub so that I can have a second view and learn the tools more in-depth? I myself am willing to give help with the English guidelines to anyone who needs it. I am thinking that the more multidisciplinarian a translator is, that is the more knowledge he/she has on all the seemingly separate translation stages, the easier it is for everyone else, since the process becomes much more flexible. If we speacialize ourselves, then we may be obstructed by situations like this one, where there are a lot of transcribers and no time-shifters or final reviewers available. With a little time-shifting tutorial by a kind volunteer, I am willing to try a trancription-timeshifting-proofreading to see if it is possible at all and if it is efficient. Would anyone else like to try it with me and exchange feedback?
Christos Pap. has attached the following file:
  • VSS Tutorial.odt (31.05 kB - 392 times downloaded - latest: Yesterday, 11:45am)
Signature from »Christos Pap.« To come to the meetings, install teamspeak from HERE and then press HERE to be transferred directly, and look for the "Linguistic Team" Channel.

2 registered users and 31 guests thanked already.

Users who thanked for this post:

Ray, Borislav/lizardman

  • "Ray" is male

Posts: 1,048

Date of registration: May 23rd 2011

Language Team: Global

Focus Group: LTI Administration Group

Location: Michigan, US

Thanks: 34585 / 6204

  • Send private message

67

Sunday, March 23rd 2014, 7:54pm

AEGISub:
Tutorials - Aegisub Manual <-- Very little on the main page, but look to the sidebar for tutorials on specific focuses.
YouTube
YouTube

Visual Sub Sync:
YouTube
Synchronizing an already made transcript [VisualSubSync]
Timing Tutorial with VisualSubSync | Jiang Hu Fansub

There are more, of course. Run a web search for "AEGISub tutorial" to find more, as needed.


Note: Also keep in mind that, on the technical side, it is imperative to always preserve the UTF-8/Unicode encoding of the files that we handle offline.
Many tools (for ex. - Subtitle Workshop) are incapable of preserving "special characters" when this technical spec is not observed/respected.

Regardless of what offline app I may use for a given function (and I generally use different apps for their relative strength in different tasks), I always finish up by opening the SRT in AEGISub, just to make sure nothing has affected the UTF-8 encoding. If it shows anything more than UTF-8, I then run it through SubtitleEdit's encoding function to return the entire file to UTF-8. If I have any need of opening a subtitle file in a text editor, I ONLY use Notepad++, keeping in mind that I have intentionally set up Notepad++' options to always use "UTF-8 without BOM". While they may not be easy to understand at first, these technical considerations are much more important than most people realize. ;)
Signature from »Ray« Earth For Sale:
Slightly Used; inquire within

33 guests thanked already.
  • "Ray" is male

Posts: 1,048

Date of registration: May 23rd 2011

Language Team: Global

Focus Group: LTI Administration Group

Location: Michigan, US

Thanks: 34585 / 6204

  • Send private message

68

Sunday, March 23rd 2014, 8:12pm

This thread is great for all of this test reporting, but we could also use a new thread focused exclusively on suggestions for the dotSUB developers on improving dotSUB to the point of making most, if not all, offline work a thing of the past. There are a couple of people showing interest in helping with LTI development needs, so I already need to contact the dS dev team to get an updated API from them (needed for our PMS automation). It would be ideal if I could also share with them some major feedback (both positive & improvement based) from our four years of efforts in their product.

Edit: After further consideration, I'll create a community thread for that after I return home from our bi-monthly supply run (time to get some food in the house).
Signature from »Ray« Earth For Sale:
Slightly Used; inquire within

1 registered user and 28 guests thanked already.

Users who thanked for this post:

Christos Pap.

© Linguistic Team International 2017
Context In Motion