How do Himalaya organise the voice acting files?

[SSH]

I've tried doing a smalling voice pack in AGS, and it was a pain. I'm also beta testing The Blackwell Legacy, which has thousands of voice lines, although probably not as many as Al Emmo. I was wondering what Himalaya do to organise all the AGS voice files and keep track of numbers, etc. It's a real pain.

Part of my reason for asking, is that CJ seems to be reworking AGS a bit at the moment, so if there's a good suggestion for improving the voice support, we could thrash it out and propose it on the AGS tech forums... so far, the simplest suggetion has just to been to allow files like EGO15_Hello_strange.ogg instead of just EGO15.ogg, which would help a bit, but I was also thinking if some kind of preview tool would be handy, perhaps playing speech back a double/triple speed so its comprehesible enough to check the wording matches, but fast enough to speed up the process. Plus auto-numbering sucks.

[GameDevChris]

I'm not sure I fully understand the question. Which part of the voice packs do you find to be troublesome -- creating an organized script for voice actors? Or simply not having a reliable and quick means of hearing which voice file is which?

Personally, I didn't have much of problem with the latter, because the sound editing program I use (Cool Edit) allows you to hear a preview of the voice files in a directory by clicking on them from the File/Open box. Although, opening Windows Explorer and going to your AGS/Speech sub-directory and then clicking on the speech files so that they play in winamp (or whatever sound program you use) works equally as well. I actually found the constant editing of scripts and re-arranging of text files/documents etc. to be the biggest and most time-consuming hassle where recording voice files (and dumping them back into the game) is concerned, and I definitely agree that this aspect of AGS could be more user-friendly.

The way we handled voices in Al Emmo was to first auto-number all of the lines, and then arrange the numbered lines by character so that each character had a script to read through from top to bottom.

The actors read directly through these numbered scripts. Since they were just straight read-throughs without any interjection from other characters, the actors were given direction on the inflection and tone each line should be spoken in. Each line was spoken about 3 times to cover all possibilities. An entire group of character lines were recorded as one big WAV file in the recording studio, and then burnt to DVD for us. We then used the program Cool Edit to split the lines and number them in the EGO1.WAV, EGO2.WAV etc. format one file at a time.  It took a long while, but wasn't too much of an issue compared to the lipsynching process.

Having recently considered the possibility of doing foreign translations of Al Emmo, a similar problem to the one you're mentioning has arisen, though. Localization teams want a script containing all game conversations with a word/line count so that they can consider how much work will be involved for their translation team.  As the developer of an AGS title, you can currently give them a neatly re-arranged document file containing all of the character's English dialogues in the same order that they are heard in the game. But the problem here is that; 1) a lot of manual editing work is required to get them all in order, and 2) The numbers of each line will no longer be chronological.  This means it would be a big job for the localization team to match the lines of their neatly organized script with the corresponsing lines in a translation template, which will be scattered all over the place. Essentially, they'd have to translate the game twice.

It would be a nice idea for CJ to implement a feature which could dump a neatly-arranged actor's script file into the game directory, which would be fully compatible with the game's translation template.  The purpose of this actor's script file would be to have a neat-looking document that you could give directly to your actors without having to edit or re-arrange it at all.  It would also be free of all of the other irrelevant gibberish that typically occupies a translation template file. It would merely be limited to containing dialogue/text from the game so that the layperson can understand it.

In the actor's script file, conversation lines would appear in the exact same order/structure as they do in the game. When being dumped, the actor's script would look at the original room files/global script etc. to determine the structure of the conversations and place them in the same order.  This would allow the actors to see the context of each sentence and know how the natural reply should flow. (Although, including an option to bunch each character's own lines together would also be a good idea, so that actors could do direct read-throughs if they wanted, like we did when recording Al Emmo's lines).

Note that the order of the lines in the actor's script file would differ from those in the translation template. Currently, the biggest issue with converting such a large game to a different language is that the translation template is useless as an actor's script, since nothing is arranged properly and it also contains a lot of junk that's not relevant to the dialogue you want recorded. It's messy. My idea would be to make it so that the translation team could simply edit the actor's script file (where all the lines are neatly ordered) by typing  their newly translated lines over/beaneath the existing English ones.

This way, if the actor's script file exists in the directory, when you click "Make translation source", AGS would first look at it and then dump all of its lines into the translation source in the relevant locations - superceding any lines that were already filled in there.

This would also be helpful for games that do not require any foreign translations, as having the ability to obtain a neatly ordered actor's script at the click of a button (and without any extra work on your part) would save a lot of time and energy.

I don't think AGS's auto-numbering feature is a problem in itself, but rather, the fact that auto-numbering seems to be useless when there's currently no reliable way in AGS to arrange dialogues/text into a useable format for voice recordings. I think the solution lies in fine-tuning the translation text/voice/dumping options, not in changing the way voices work.  It's a given that doing a voice pack with thousands of lines is going to require a lot of time and effort and there's probably no way you can speed through that process. But it could be made as hassle-free as possible. I think the suggested approach would solve that problem.