DIY Voice Pack CreationNote: These instructions are for the original tool, not the 3rd party app created by AbsolutePhoenix
Required tools:
- DBVO Voice Pack Maker - Go to the DBVO Main Mod Page, join the Discord, and download it under the tool-download channel.
- In the same tool-download channel, scroll up and grab the FonixData.cdf file as well.
- zEdit - from Github releases page
- zEdit dialogue extraction script - found on the DBVO Discord under the tool-guides channel, scroll up a bit and you'll see it pasted as plaintext that's downloadable. Drop this into your zEdit/scripts folder. You can search for "topics_to_txt_0.0.1." to help you find it.
- Yakitori Audio Converter this one is hosted on the Skyrim page, there are three different pages for this on Nexus and it does not matter which you pick.
- This will now also require at least the lowest tier of subscription account, which is $1 for your first month if you've never signed up before.
Topic Extraction Process:
- Download and install the mod you want to create a voice pack for.
- Launch zEdit.exe and follow the instructions to add Skyrim Special Edition.
- Click "Start Session" and select the esm/esp file you would like to extract the topics from.
- Once zEdit has loaded, right-click in the left view and select "Automate."
- In the list, find and select "topics_to_txt_0.0.1."
- Change the "pluginName" field to the plugin you want to extract topics from.
- Update the "outPath" field to the desired save location for the topics.txt file (use double backslashes "\" in the path).
- Click "OK" and wait for the extraction process to complete.
First time pack creation will require you to clone your own version of Bella (**only possible on subscription accounts**):
- Go to the files section and download the Bella_Voice_Clone_Source from "Update Files", the zip contains a couple of 1m45s long spliced dialogue clips.
- Log into elevenlabs.io, go to the "Voices" section and click "Add Generative or Cloned Voice".
- Select "Instant Voice Cloning"
- In the name field put "Bella"
- Drag the spliced dialogue clips to upload, tick the box at the bottom and click Add Voice.
- Bella (cloned) will now be in your voice library.
- You can close the website as the rest will involve the Voice Pack Maker.
- This only needs to be done once, or for as many API keys as you have access to.
- Highly recommended to use VPN for multiple API keys as ElevenLabs does attempt to track that and could flag and/or ban your account(s).
Voice Pack Generation Process:
- Launch DBVO Voice Pack Maker. (DBVO VPM)
- Paste your API key into the top box and click apply. (Refer to main Sticky Post to see how to retrieve this)
- Select Bella(cloned) from the second drop down.
- Below that are sliders to adjust Stability and Clarity, by default they are set to 75. I have NEVER touched these.
- Below the sliders in the next drop down, Select Eleven Multilingual V2 or Eleven English V1. IMO Multilingual V2 has better emotive speech and more correctly intuits the tone of the lines being fed. It is the selection I use for all patches created 3/13/24 and beyond.
- At bottom left, hit select Topic Files
- Navigate to the folder you saved your extracted dialogue topics file to and select it.
- Hit the SELECT FONIXDATA.CDF button and navigate to wherever you saved that file earlier.
- Hit the Select Output Folder and tell DBVO VPM where to save the files.
- Make sure Generate LIP is checked (should be by default), then press Start Generating.
- Once generation completes, go to the output folder you specified earlier. You'll see a bunch of .WAV and .LIP files.
- Open Yakitori Audio Converter. Set Input format to .wav, Output format to .fuz and make sure mode is set to Lip required
- Drag and drop all of the files from the output folder to Yakitori and press Convert. Close Yakitori unless it printed errors in the bottom right window, if it did you will have to address whatever they may be.
- Go back to DBVO VPM, click the second tab at the top nex tto ElevenLabs Batch Tool, "Voice Pack Maker"
- For display name, enter "bella" without quotes. For unique ID enter "voicebella" without quotes. It is important to match these exactly so the pack is loaded when you select Bella in DBVO in game.
- At the bottom click Select Folder Containing FUZ Files, your previous output folder should already be selected so you can just hit Select Folder right away.
- Click Select Output Folder (This will contain the Voice Pack Ready to Ship) and choose a (preferably different) folder to store the completed pack.
- Click Create Voice Pack
- You're DONE!
- Go to the completed pack folder you just selected and add the DragonbornVoiceOver and Sound folders to a zip file, then drag and drop that into the downloads for MO2. Install as normal and activate.
Tips for overall process:
- I use a multi-folder setup. Inside the folder where I installed DBVO VPM, I have a folder for Topic files (which is where I tell the zEdit script to dump them), a folder for working (where I tell DBVO VPM to dump the generated .WAV and .LIP files), and a folder for completed packs (where I tell DBVO VPM to put the finished product.
- Once you've extracted your dialogue topics, use a website to check the character count (I personally use https://wordcounter.net/character-count). If it is over 30,000 you will need to process your patch over multiple months at the $5 Tier, or upgrade your account to creator.
- If the dialogue topics have lines containing <Alias=Player> those will have to be generated using whatever player name you intend to use.