Apparently, I'm one of only a few individuals who still has access to the CSLU Toolkit, software that can output phonemes based on an audio and text sample. Historically, this has been pretty much the only way to generate LIP files for KOTOR so characters' lips will flap when they talk. LipSynchEditor converts from CLSU's PHN format to the LIP format, but to do that you need the PHNs first... unless you want to make lips manually, but that isn't practical. Unfortunately, the Center for Spoken Language Understanding's servers have been offline for a long time. As such, it's currently impossible for new users to install the software required to generate these PHN files. Only a few veterans like myself still have it installed. And I can't even guarantee I'll have access to it forever. It's currently installed on my old computer, but not on my newer system that I built after the servers were offline. I doubt the computer that does have it will last forever, so it's probably only a matter of time before I lose access to CSLU too. Until then, though, I'm able and willing to generate lips on request. And I've gotten a few requests already.
Now, I don't want to make a habit of this and get stuck as the guy who does everybody's lips for them because making lips with the CSLU Toolkit is a long, boring process. What I can do for you, though, is tell you can do for me so I don't have to do all of it for you. If you follow the instructions below and send me all the necessary files, all I have to do is hit a button and send stuff back to you. And I have no problem doing that.
- CSLU requires three things: 1) audio of everything you want to create a lip saved as an individual mono WAV files; 2) a text file accompanying each audio file that contains the words spoken in that line of dialogue; 3) a master list matching each audio file to each text file so it knows what to process.
- First, you need your VO files in the mono WAV format that the CSLU Toolkit wants. If your VO is new, you merely have to make sure to save it in this format. If you're working with VO from the game, you'll need to convert it first. The game VO files are generally MP3 files with extra bytes added at the start of the file to confuse us. Stripping this header will restore them to regular MP3 files. This can be done with a batch script and you can read up on that here. That will make them MP3, but they still need to be converted to WAV. Any old audio converter can do this. Freemake Audio Converter is one I use for Windows. Alternatively, you can use the Miles Sound Tools to play and convert the VO directly from the original game format to mono WAV.
- Next, you need to write out all the words spoken in every line and save them as text files. I typically name the text files the same as the audio files. You then need a list of all your audio and text files. This is the most tedious part of the process, but fortunately I've attached batch scripts below to make this easier.
- text.bat will create blank text files for each audio file. You still have to type the dialogue into them, but at least you won't have to worry about the file names.
- list.bat will create the master list. It will search for every WAV file and create a line in the list for each one, saving all this as lips.txt. My script assumes each text file is named the same as each audio file, so if you want to use this script, that's required.
- Send me: 1) all your mono WAV files; 2) all your transcribed text files; 3) your master list of everything to process. Once I have all of those, I can run CSLU's script to generate PHNs and send them along. I can also batch convert to LIP if you want, because that isn't nearly as big a deal as all the above.
- If you want me to make lips files for you, post a comment below with the necessary files attached or linked, and I'll get to work.