Skywind Filecutting Guide
Filecutting Tools
Toolkit for LibreOffice and Audacity
The wonderful Krisscracka and Slaynot have developed a toolset to make filecutting easier.
The latest release can be found here: https://gitlab.com/Slaynot/filecutter-toolkit/-/releases/1.4
Information on installing and using the toolkits can be found here: https://gitlab.com/Slaynot/filecutter-toolkit/-/wikis/home
Journeyman Toolkit
The Journeyman Toolkit from MiGeek handles formatting of Retakes scripts, updating labels for changes in filenames, and marking missing and unused lines on cut files.
The latest releases are currently posted to Trello: https://trello.com/c/N3mxgZL2/
Prerequisites
For filecutting, you will need two pieces of software:
Audacity, a free audio-editing program. You can download it here: https://www.audacityteam.org/
A spreadsheet editor, such as Microsoft Excel or OpenOffice Calc. Excel is expensive, so if you don't already own it we recommend open-source software like OpenOffice or LibreOffice. We recommend LibreOffice in particular (even if you have Excel) because of the toolkit linked above---this will make filecutting much easier!
Additionally, you will need:
A Trello account. This is free, and you can register for it here: https://trello.com/. You will be given an invite to the Skywind VA board once you've completed training.
A Discord account. Make sure you are on the Skywind server: https://discord.gg/skywind
What is filecutting?
When voice actors submit their recorded lines, the lines are all in one big file. The job of the filecutter is to take this big audio file and create a "label track" that assigns a filename to each segment of the audio submission. You won't actually be creating the individual files yourself --- that will be done automatically using your label track after our SFX department has mastered the track. In addition to assigning filenames, you will check for pronunciation and script accuracy, and decide what line to use if a VA submits two takes of the same line.
So, your task will be to download a VA's submission and script, create this label track, and then create a document of any lines that were missing or need to be redone.
The only requirement is that when you are given a clip that you try to get it done. If you can't, get as far as you can, tell us where you stopped, and send what you've completed so someone else can finish it off. Try to do about 30-50 cuts at a time so you don't burn yourself out. Remember to do it from front to back so to make it easy to keep track (this hopefully naturally would be what you would do anyways).
Each script will be around 500 lines, with the biggest ones containing 2000 lines. We do understand that the larger cards can be intimidating and difficult to tackle; if you want, feel free to ask to only cut a segment of these cards, like a set of 400 lines. Don't burn yourself out!
What do scripts look like?
Below (Figure 1) is an example of a Skywind script. The first row contains information about the character(s) being voiced. Things like names, race, etc... After that is some useful links, and then we have the script itself, where Row 3 is the headers and Row 4 and onward are the VA's lines.
Column 1 is the filename for this segment. This is what you will label segments as.
Column 2 is the speaking NPC.
Column 3 is what the player asks about or does to get this NPC response. This may be blank for things related combat, detection, or idle NPC behavior.
Column 4 is the line the Voice Actor reads.
Column 5 is the facial expression the NPC will have in game, and also gives you a tip about the emotion the VA should be using. Note that these are not always very descriptive, so if the VA reads the line well but it doesn't perfectly match the Facial Emotion, then that's okay. If Acting Notes ever conflict with the Facial Emotion, go with the Acting Notes.
Column 6 is any acting notes for this segment. This includes pronunciation guidelines.
Column 7 is where you will leave notes if the line is missing or needs to be redone!
Instructions
1. Download the files of your assignment (lines and matching script) and open them up in Audacity and your spreadsheet editor respectively. When opening in Audacity, use the "File -> Import Audio" function. These files will be attached to the VA's card on Trello. If you can't view the attachment, let the filecutting lead know.
2. Verify that the audio was recorded in 44100Hz, 16-bit WAV file (see Figure 2).
If the VA recorded in a HIGHER sample rate or bit rate (e.g., 96000Hz or 32bit), then this is fine and you can proceed like normal. Our sound mastering team will take care of this down the road.
If the sample rate or bit rate are LOWER, or the file is not a WAV file, then let the leads know and do not proceed. After the leads get back to you, they will tell you to either go ahead and cut, or have the VA redo their work.
Figure 2 - Proper format.
3. Work through the audio from start to finish. You will want to zoom in to see each line clearly. For each line read in the audio recording:
3.1. Find the row of the script that this audio segment corresponds to.
If you can't find the audio segment anywhere in the script, then we do not need that line, and you can skip it. Just please triple-check that the line really isn't in the script. See the "Tips and Reminders" section for more info.
3.2. Listen to the segment, and look for the following:
Does the VA read the line correctly?
Is everything pronounced right? (Both Dunmeri words and English words)
Is the acting good?
Is the sound quality good? (only look for glaring issues)
Is the line free of typos and grammatical errors that would require a retake from the VA? (More on this below.)
If the answer to the above 5 are "yes", then assign a label (Step 3.3). If not, then mark down what the problem was (Step 3.4)
Note: Sometimes VA's will do multiple takes of the same line. If this happens, pick the one you think is best, and only label that line. Or, if you find a line that has a mistake, verify that there isn't a second take right after it, before marking the line as needing to be redone.
Note: Regarding grammatical errors:
We aren't looking for perfect grammar but rather grammar that reflects an ordinary speaker. Ordinary speakers will mix up good vs. well, drop prepositions or "that"s from clauses, start sentences with conjunctions, use double negatives and slang, etc... Morrowind dialogue frequently uses this, and we don't want or need to change that.
Morrowind also frequently uses some archaic word meanings and phrases. Again, this is okay, and is a reflection of Morrowind's setting. If you aren't sure if this might apply to a line, just ask!
If you see a grammar error that would require a retake, please ask about it on the Discord to confirm whether the line really does need to change. Changing a line can affect a large number of VAs so we need to be very careful about what we change.
3.3. Assign a label:
Highlight the segment in Audacity.
Include a half-second buffer of silence before and after the line.
This buffer gives our audio engineers a good measure of the room tone/noise level during the line.
Lines can have less silence: 0.3 seconds before an after is OK, and 0.1 seconds is the bare minimum. If there's less than that, we have to consider the line as having sound quality issues, so proceed to Step 3.4.
The easiest way to do this is by using the "Length and Center" option for audio position, and adding 1 second to the length. See Figure 3 for an example.
DO NOT add silence if there's not enough there.
Add a label by pressing Ctrl+B. The first time you do this, Audacity will create a label track for you. The following times, labels will be added to it existing track.
Use the filename of the segment for the label. If you copy and paste the whole cell from the script spreadsheet, make sure to delete the space that the Audacity adds at the end.
Once you've added the label in Audacity, go back to the script and highlight that line with the color CYAN.
Note: See the "Filecutting Tools" section for a Macro you can use to make this easier.
Figure 3 - Example of expanding selection to include silence.
Note: Sometimes a line will be duplicated in the script, but with different filenames, as in Figure 4 below. This is not a mistake! You can simply assign the multiple filenames to the same segment of audio, as shown.
Note: It is perfectly okay for lines to share silence, as shown in Figure 5 below.
Figure 4 - Duplicate Lines with Different Filenames. Make a label for each filename!
Figure 5 - It is okay to share silence between lines.
3.4. OR: Fill out an error report on the line.
Go to the script, and highlight the row according to what the error was. See Figure 6 for what color to use.
If the problem is a pronunciation or script problem, highlight the specific word or phrase in RED.
Add a note in the "Filecutter Notes" column about what the error was. Try to be brief but descriptive and helpful.
Remember to keep your comments friendly! We don't want to discourage VAs about their lines. Avoid saying things like "Bad Acting".
Note: Black lines are a special case right now! Do not apply this color yourself. See Temporary Important Notice! for more details.
Note: If you use LibreOffice, see the "Filecutting Tools" section for a macro to make filling out the Error Report easier.
Figure 6 - Error Report Color Coding
4. All done? These are the final steps to finish up your task:
4.1. Export the Audacity label track (a .txt file), as in Figure 7. The label track should be named:
Labels_<the exact name of the audio file>.txt.
4.2. Save your script, and create a zip file with the script and label track. The zip file should be named:
Ready_For_Checking_<Main or Retake Number>_<VA Name>_<VA Race>.zip
4.3. Upload the zip file to the Trello card, then move the card from to the "CUT RECORDINGS" column. (Or "CUT RETAKES" if it is a retakes card.)
4.4. Ping the filecutting lead and journeyman filecutters on Discord when you're done.
4.5. Give yourself a big thumbs up because file cutting is hard work and you're the best!
Note: More details on the naming convention for label tracks and zip files can be found here: Naming Convention
Figure 7 - Exporting Label Tracks. This label track would be named Labels_Vodunius_Missing_Lines.txt.
Tips/Reminders
Remember: Do not delete or edit any audio, only label segments using Audacity's labeling tool.
Please leave friendly and helpful acting notes for our voice talent. You are their acting coach! A good sensible judgement of acting value and skill is needed for this. Use you're best judgement, and feel free to ask other filecutters for a second opinion. Quality is very important to us here at Skywind!
Trouble with shouting lines? Don't worry you are not alone! Be sure to look to the label name itself and the mood of the character in the script for hints! For example, the line for "MorroDefau_MDQPowerAttack_0004A0F7_1" has no prompt, but the filename tells you that it is for when the NPC is power attacking.
Don't be afraid to tell things as they are. You are the last bastion of quality after all! If a VA is not convincing enough, report it! This is especially important for combat lines.
If you are unsure about a pronunciation and don't see it on the guide, feel free to ask on Discord!
If you have any questions at all, feel free to ask on Discord!
What Next?
After you've cut a few cards, you should be ready for "Journeyman Filecutting." Basically, this is doing a second-pass on cards to review the work of another filecutter. Measure twice, cut once, as they say. ;) If you are interested, you can check out the guide for this here: Journeyman Guide. Once you've reviewed the guide, get in touch with a lead to get a second-pass assignment.
Old Script Historical Info
As of November 2019, we started using a new tool to generate voice actor scripts. There are two things about this you need to know:
The new scripts simplify how each segment's filename is presented: Each row has one segment and one filename. The filename for that row corresponds to the segment for that row. Simple!
Dialogue that was previously planned but not implemented, and only had a placeholder name, has all been put in game and given a real filename.
For posterity, the instructions for cutting these scripts can be found here: File Cutting Instructions (Old)