How to make mandarin karaoke in SSA?

[h3170ra]

Hi, I already succeed make mandarin subtitle but when I want to use the subtitle filter,
it doesn't work!!! I just read the thread in the forum and find out that that filter doesn't support unicode.
Is there some that can tell me where I can find filter that work in virtualdub and support unicode?
Oh yeah, btw I've watched gundam seed that subbed by fans, I saw how they make the karaoke, it's really cool.
It can pop up or change color. I tried to make it using substation alpha, but when I previewed it in virtualdub,
the karaoke effect doesn't show up. I read that I have to use textsub filter. Where I can download it?
Btw is the textsub filter support unicode?

Thanks a lot. :)

Best Regards,

[Phanton_13]

Vobsub ( VSfilter )suport unicode, textsub is a part of this filters, but im not sure but can be only utf16 format. it can be found at: http://sourceforge.net/projects/guliverkli/ , rename the vsfilter.dll into vsfilter.vdf and put in virtualdub pluiing directory

[h3170ra]

Thanks a lot, man. I tried ur suggestion and it works. I really appreciate it.
Bout the karaoke effect, I've browsed and found out that I have to use ASS not SSA, with external effect ( After Effects, etc... ).
But since I only newbies, not an expert. I don't know how I gonna do that? What subtitling program that can do this and support unicode?
I'm complete blind about this. Could someone help me by giving me step by step guidance?

Thanks in advance. :D

[Liisachan]

Altho .ass is more flexible, you can use normal SSA for Karaoke too.
Unicode is recommended (there are some subbing tools which support Unicode like Medusa) but you can use non-unicode code pages too, if you speficy the Encoding like

Style: style1,MingLiU,30,65535,10015224,0,3355443,-1,0,1,2,1,2,30,30,30,0,136

... where 136 means Traditional Chinese, and 134 would mean Simplified Chinese. I made a small Karaoke sample in Chinese (http://park14.wakwak.com/~flower/falafala.zip) with non-Unicode SSA, generated by Sub Station Alpha and a text editor, hardsubbed with textsub.vdf (sry, this is traditional Chinese, because the song is from Taiwan)

Since I don't know Chinese, timing may be wrong, and/or maybe I made silly mistakes in Chinese language anyway :D but I hope you can get general idea from this smaple.
If you know how to karaoke in English, you might want to Karaoke using Pinyin first, then you can edit your SSA with your editor, replacing Pinyin into Chinese characters.

[h3170ra]

Thanks Liisachan for ur suggestion, :D

Too bad I can't download ur sample. The file size is too big 4 me.
Since the internet connection in my country is very slow. I'm from Indonesia. :(
Right now I can do the karaoke effect using substation alpha,
but the color changing word by word not letter by letter.
Liisachan, do u know progr that can do this(letter by letter?
Do you know how to make karaoke effect that I saw in some anime movies?
The effect like pop up, etc? It's really cool if I can do that.
Bout medusa, I've tried to download it. But it always get disconnected. :(
Btw u r a programmer?

Thanks in advance.

Best regards,

[Liisachan]

What you want to do is possible with normal SSA too.
Sample SSA (http://park14.wakwak.com/~flower/falafala-ssa.zip) (2kb)
Download it, load it onto VirtualDubMod using Textsub.vdf, with any video clip (for testing), and use arrow keys to check frame to frame to see what (and how) exactly Karaoke is. You might want to check this small portion of vid (http://park14.wakwak.com/~flower/falafala1.zip) (479kb)
Then, open that SSA with your editor such as Notepad, and get a general idea what you should write to realize karaoke.

Originally posted by h3170ra
Right now I can do the karaoke effect using substation alpha,
but the color changing word by word not letter by letter.


Read docs in \Gabest\VobSub\Docs and check that real sample. Use [\K} for gradual change, use {\k} for immediate change, both can work letter by letter. I put both \K and \k in my sample so you can observe the difference...

Like I said, I made that sample with Sub Station Alpha and a plain text editor. Any timing tool (+text editor) should do. Karaoke is easy. Essentially, you just have to time for each letter (if you want to change color letter-to-letter). For instance, if the duration for A is 30 centiseconds and the duration for B is 40 centiseconds, type
{\K30}A{\K40}B

[h3170ra]

Wow, thank you, arigatou, terima kasih, xie xie, liisachan
Now I can do karaoke letter by letter.
Btw I have tried the lead time function in substation alpha,
but it seems doesn't work to me.
I already set the lead time, even to 1000 seconds!!!
But the text still not appear earlier.
Can you tell me What is wrong?

Thanks in advance.

Best regards,

[Liisachan]

What do you mean by 'the lead time function'?

Somoething like this?

;exact timing
Dialogue: Marked=0,0:00:10.00,0:00:20.00,style1,,0000,0000,0000,,{\K80}Kara...

;preloaded
Dialogue: Marked=0,0:00:09.00,0:00:20.00,style1,,0000,0000,0000,,{\k100}{\K80}Kara...

[h3170ra]

Yes, something like that. Is it possible to do that automatically not manually edit the ssa using notepad?

Thanks in advance.

Best regards,

[Liisachan]

Basically you can use Sub Station Alpha: Timing -> Shift Times... in the menu

[h3170ra]

But if I do that, the karaoke will appear earlier too.
I don't want that happens. I only want to have the text appear earlier. :)


Thanks in advance

Best regards,

[Liisachan]

for instance, shift the Start Time 1.00 sec backward (earlier), then use your text editor to replace every 0000,0000,0000,, in Karaoke lines with 0000,0000,0000,,{\k100}

Did you notice the void {\k20} at the start of each line in my sample file?

Dialogue: Marked=0,0:00:08.06,0:00:10.91,style1,,0000,0000,0000,,{\k20}{\K59}Hu...

The point is, the above line is not equal to this:

Dialogue: Marked=0,0:00:08.06 ,0:00:10.91,style1,,0000,0000,0000,,{\K79}Hu...

even tho 20+59 = 79

The extact timing was,
Dialogue: Marked=0,0:00:08.26 ,0:00:10.91,style1,,0000,0000,0000,,{\K59}Hu...

That line will be preloaded by 0.2 sec, and karaoke color-change begins 0.2 sec leter. So the Start time is now 0:00:08.06 and we will wait for 0.2 sec by that void {\k20}

In the real work, one might do a bit more complicated thing like
Dialogue: 0,0:00:10.00,0:00:20.00,style1,song,0000,0000,0000,,{\move(320,20,320,0)}{\fad(0,500)}{\k100}{\K20}foo{\K30}bar
but the basic idea is the same.

[h3170ra]

ic...
I get it now. So I have to do it manually.

Thanks a lot

Best Regards

[Liisachan]

What you'll have to do is basically simple (just shift the Start time ** centisec and add {\k**} for each line), so you can do it with a small Perl script which you could write in 5 minutes or so. Or maybe I'd make a win32 gui application for you, if you can wait a few days (since im currently a bit busy)...
Altho lots of manual, delicate tweaks are needed for HQ karaoke, decent karaoke effects are not that difficult to realize. Anyways you cannot learn just by reading. Try and error--write SSA/ASS a lot, chcek the outcome in VirtualDubMode etc. :)

you have 2 different problems here:
(1) How to time for karaoke and how to tweak the timing
(2) How to use Unicode in SSA, or if not Unicode, how to control the code page.

[ak_gamma]

It's almost too easy.
For quality karaoke timing -use such tool as XombieSub for example. And Unicode in ssa is trivial: you just metion the encoding you need:
0 for English
128 for Japanese
...
And writing ssa/ass code - you're quite right - it's the best way to do quality and attractive karaoke.
wanna my example - try this video: www.pgsm-sub.fatal.ru/op.rar (1100 kilobytes, RAR 3.20)
it is encoded using On2 VP61 codec.
You can see, that besides ordinary karaoke \k and \kf there're also a lot of extra features, but it's rather easy to write. Of course, I used ASS, but using textsub for hardsubbing -there's no difference: ass or ssa :))

[Liisachan]

A bit off topic, but yeah, fancy karaoke is so common nowadays. Some of them are amazing and personally I like those eye candies.

One thing I dont really like is, altho they are beautiful superficially, romaji and/or translation is incorrect as often as not. I'd prefer accurate/good translation with simple karaoke to mistranslation with showy karaoke effects. (In your sample, AI BIIMU is mistranslated as 'love beam' tho it should be 'EYE BEAM'; Romaji for JU is not JYU; Not "Try get a chance" but "TRY GET CHANCE"--cant be helped, so called engrish ^^; )
That said, I'll post a link to another fancy sample (http://www.geocities.co.jp/MusicHall/2722/ddkara.zip). I hope
ak_gamma's sample and this one of mine will inspire ppl who are intersted in karaoke in ssa.

[ak_gamma]

Liisachan
I'm 100% for CORRECT subs and no other way.
that karaoke has the only mistake - "Romaji for JU is not JYU"
That translation karaoke wasn't done by myself - I was only to add some effects.
Not long ago, I was given a task for karaoke this song once more, and the transcript was completely different - close to what you said.

"Hitomi wa itsumo jueru.
My eyes are always jewels.

Egao yori kuchibiru yori aibiimu ga kimewaza.
Instead of a smile or lips an eyebeam is my special weapon. "

But it is still TRY get A chance - it's written in the oficial japan text of this song - so there's NO OTHER WAY :))))

I took a look at your karaoke - funny, but we both still use ASS features. not ssa, ssa has NO fade :)) as I can remember.

Do you have some more examples, I wanna see and try to remake it. the link you gave - is clear enough how to do it :))

[Liisachan]

Sorry, ak_gamma,
...since the audio in your clip is very low-quality (24 kbps), I cannot get every words for sure; I might have been wrong. I just wanted to say, generally, plain but accurate karaoke is better than showy but inaccurate subs.

Getting lyrics is always difficult, especially when the singer who is not english-speaking sings something in quasi-english... it's not your fault :D Even famous fansub teams are sometimes having trouble with OP/ED. altho many teams are doing great jobs too. Well, another typo in your clip is "Maiku appu" which should have been "Meiku appu." just fyi ;)

Now, back to the main topic...
We want to sub in MultiByte Chars, like Cantonese, here.
you said using unicode is trivial, but things are a bit complicated...

1) If the SSA is in UTF-16 LE with BOM (WCHAR), then textsub.vdf can handle it as it is, even if you don't speficy the 'encoding' parameter in the style definition. Even if you use 0 instead of 128 for Japanese, textsub.vdf/vsfilter.dll can handle the encoding correctly.

2) Textsub.vdf doesnt like a file in UTF-8, while mkvmerge is ok with UTF-8 as long as it has BOM.

3) If the file is in MultiByte Chars, not in Unicode, you _must_ specify the encoding (like Japanese Shift_JIS = 128) as you said, even for hardsubbing.

4) When softsubbing, you _should_ specify that encoding param. Win98 will need it. (Win2000/XP doesn't need it)

5) On linux, that parameter will be generally ignored. The same is true with the {\fe} tags. So you should not use {\fe} in softsubbing.

[ak_gamma]

Liisachan
as I have already said - for the video, you've seen - I di only effects!
I redid that karaoke for some new soundtrack and the lyrics WAS different enough - so "Meiku appu" is correct there.
The lyt\ris of karaoke you saw, was made by another fansub group - TV-Nihon.
About poor audio quality - the reason was size of file. And about the karaoke itself, I always do it plain - with the features ass provides - \k and \kf (maybe \ko, but haven't done yet)
And I think as you, that QUALITY of translation is the main goal and the look isn't so important.
About Uniode - I said it's easy - cause you said yourself - with Unicode it's REALLY easy.

offtopic- do you really from Japan? Where're you from, I wonder.
What karaoke timing prog do you use? I use XombieSub for example