So here's what I've been able to figure out regarding Point and Click audio:
-For some reason they need to be over written through a different folder chain (\Data\Public\Shared\Assets\Sound\ according to this mod https://www.nexusmods.com/baldursgate3/mods/3626?tab=files) -Unfortunately this version of the files are 10 digit strings of only numbers with no discernible pattern I can figure out.
However, looking in to the .bnk files (which I've learned are merged libraries of multiple .wem files?) I found one labeled Vocals and was able to convert it to .xml. I found a chunk of code containing the 10 number string of a Tav1 file I identified by listening to a few random .wems. The 10 digit string was listed as a "sourceID", above it in that same chunk was a 6 number string labeled ulPluginID. When I searched for more occurrences of that string, the few corresponding sourceIDs I looked at played as other Tav1 click lines.
So if someone has any ideas on how to more efficiently pull those number strings from this file I would be SUPER grateful
The lsf file is broken into code chunks named after the category of click line (i.e BGClickReaction_ReptAction_PortraitClickDowned)
For each of these chunks there is a line labeled SoundEventID, you'll want to copy the code listed as its value.
Now you will move to the .xml file
Search for the SoundEventID in the .xml, there will a single obj na="CAkEvent" section with that number listed as its ulID, then copy the corresponding ulActionID value.
Searching for the ulActionID will identify a single obj na="CAkActionPlay" listing the value as its ulID, copy its corresponding idExt
This is where it becomes a bit more complicated
Searching for the idExt will find multiple (16 at least for what I've checked so far) obj na="CAkSwitchCntr" listing it as their DirectParentID. This seems to be the separation between origin characters/Tav voices as far I can tell, I need to do some more checks to figure out if there's any pattern in how their listed.
From any of these you can scroll down a bit to find a obj na="Children" and copy the ulChildID (For categories that have variants for Combat and Stealth it looks like there will be a separate Child for each) Searching for this will bring up multiple obj na="CAkSound" with that number as their DirectParentID. The corresponding sourceID is the name of a .wem containing a voice line of the initial category spoken by the specific character assigned by CAKSwitchCntr (at least from what I've seen thus far, I'm not quite sure how this accounts for all of the Tav voice options)
I recognize this is a lot, so hopefully I or someone else can figure out a way to automate some of this process
I'm super appreciative for your starting this as well! Also I don't know if you're familiar but the Bg3 dialogue parser (https://drive.google.com/drive/folders/1GR3CjFtM3u3-V5KrJkemTsAhx6VRUShy) might help people in tracking down audio files (sadly only seems to pull Tav 1 voice?)
Thank you for sharing this! This is all very new to me but I am determined to change the point and click voice for my character. Where can I find the file
so that I can extract it? I'm not finding it in the game folders
Edit: Figured out that you get the file path from the Shared.pak file and extracting that. Now I'm trying to figure out the next part with the ulActionID
Also, I wanted to ask where are the .wem files located? The ones that correspond to the sourceID names? Edit: I am going through the SharedSounds Sounds folder and it seems like it would be those .wem files in there.
Did you ever get the replacement done? I’ve identified the SourceIDs for the point and click audio but I’m not sure where to put the new recorded wems to overwrite them
hey there, i made a video tutorial for tepig327's instructions along with a program my husband and i made for speeding up some of the process! i recommend watching on youtube as everything you need is linked in the description!
This thread has been incredibly helpful. Does anyone know where to put the rerecorded point & click voicelines once they're finished? Do they just go in with the other lines in the /soundbanks folder, or do they go somewhere else?
Hi, I just finished this tool: Baldur's Gate 3 Dialogue Finder (BG3DF). It allows you to search all in-game .wems (including point-and-click) by dialogue, character and file name. Hopefully it'll help to streamline a few of the processes you guys have been going through.
My most heartfelt thanks for all of the research and work everyone in this chat has done - this is an extension of it and I couldn't have made the tool without it.
Holy cow! I'm so sorry I sort of fell off the face of the earth after posting what I found. I can't believe you were able to compile it all like that!!!!!
Anybody could help me with my wwise using issue-which when i convert wav to wem,my wwise just create multiple folders like 00 0A 0B otc. with new code named files,instead of windows folder with original name files.Why is that?Heeeeeelp me plz~~
I had the same issue and ended up also installing an earlier wwise version. I installed and used 2021.1.0 and after converting, it created the Windows folder as expected.
Hi guys. Tell me please. I did everything according to the guide. I'll give you an example. I have the voice of Kratos, which is tied to the 3 voices of TAV, and in order for it to work as a point and click, I have to find ten-digit files, listen and compare them with the files that I recompiled according to the guide, and rename the recompiled ones. Do I need to listen to all 4 thousand files?! T_T
Can anyone explain why even guide files don't give any effect to the game sounds? I did all according to youtube guide for my .wem (Vorbis) files and it didn't work. I've decided to check any other files and it's not working too (like delete some voicelines from the game). Has anyone run into this?
Have you ever had a file you want to replace, let’s say 826541967.wem, just not pan out no matter what you try? The original has stereo audio at 44100Hz. I cut the new audio to match in Audacity, export as .wav, convert in Wwise to Vorbis with matching settings as the source, and no luck. It cuts out the audio in game, but does not replace it, though the exported file can be played in foobar2000 no problem.
I use this file as an example because there is an existing mod using it, and I thought, if this mod works, it’s not cutting out because I can’t use audio replacers. If I use the original mod, the sound is replaced just fine, but even if I take their .wem, convert it to .wav, then back to .wem, it no longer plays in game.
I have been going through trial and error for days trying to figure this out. I’ve posted to the BG3Modding Reddit as well as the Discord, but no one seems to know what is going wrong. Here's a screenshot of my settings in Wwise when converting.
I’ve tried WEM Opus just for giggles. I’ve tried using different sample rates on purpose, using versions 2022, 2023, and 2024 of Wwise, installing Wwise on a different machine and converting there. All no luck. I must be over looking something. I’ve followed the Replacing in-game Sound instructions from the Modding Community Wiki, read Custom Battle Music Audio Replacement on Nexus, and watched the Audio and Voice Replacer Guide. I just don’t know what else to do. I’ve done model edits and it’s so easy, I must be missing something simple!
MY BIGGEST AND MOST HEARTFELT GRATITUDE TO @SIMOSAS1 FOR ALL THE HELP HE'S PROVIDED My durge voice 4 has been effectively exorcized of the new patch7 maniacal laughters and spoilery lines thanks to his wisdom and kind attention! While still being a durge voice, with its occasional outbursts of creepiness, it does no longer sound deranged and out of control, just the way it was pre-patch 7. This was achieved by introducing blank sound files as replacements for the following: 525773031.wem & 601217730.wem = maniacal laughs 185320340.wem = father, they will die for you 274409118.wem = father, are you watching 134440127.wem = they are dying for me. all of them (this one was not spoilery but the high pitch on the "me" sounded so much like deranged Orin it just got on my nerves). Files can be created as .txt files and just plainly change the extension to .wem They should be placed in your Data\Public\Shared\Assets\Sound folder. The result should be the aforementioned voicelines not playing any more. THANK YOU SO MUCH AGAIN!
58 comments
-For some reason they need to be over written through a different folder chain (\Data\Public\Shared\Assets\Sound\ according to this mod https://www.nexusmods.com/baldursgate3/mods/3626?tab=files)
-Unfortunately this version of the files are 10 digit strings of only numbers with no discernible pattern I can figure out.
However, looking in to the .bnk files (which I've learned are merged libraries of multiple .wem files?) I found one labeled Vocals and was able to convert it to .xml. I found a chunk of code containing the 10 number string of a Tav1 file I identified by listening to a few random .wems. The 10 digit string was listed as a "sourceID", above it in that same chunk was a 6 number string labeled ulPluginID. When I searched for more occurrences of that string, the few corresponding sourceIDs I looked at played as other Tav1 click lines.
So if someone has any ideas on how to more efficiently pull those number strings from this file I would be SUPER grateful
Following a slightly altered version of this tutorial: https://github.com/HumansDoNotWantImmortality/bg3-sids/wiki
It requires alternating between two files:
Extract Public/Shared/Assets/Sound/VOCALS.bnk and convert to xml
Extract Shared\Public\Shared\Content\Assets\Sounds\Vocals\[PAK]_Vocals\merged.lsf
The lsf file is broken into code chunks named after the category of click line (i.e BGClickReaction_ReptAction_PortraitClickDowned)
For each of these chunks there is a line labeled SoundEventID, you'll want to copy the code listed as its value.
Now you will move to the .xml file
Search for the SoundEventID in the .xml, there will a single obj na="CAkEvent" section with that number listed as its ulID, then copy the corresponding ulActionID value.
Searching for the ulActionID will identify a single obj na="CAkActionPlay" listing the value as its ulID, copy its corresponding idExt
This is where it becomes a bit more complicated
Searching for the idExt will find multiple (16 at least for what I've checked so far) obj na="CAkSwitchCntr" listing it as their DirectParentID. This seems to be the separation between origin characters/Tav voices as far I can tell, I need to do some more checks to figure out if there's any pattern in how their listed.
From any of these you can scroll down a bit to find a obj na="Children" and copy the ulChildID
(For categories that have variants for Combat and Stealth it looks like there will be a separate Child for each)
Searching for this will bring up multiple obj na="CAkSound" with that number as their DirectParentID. The corresponding sourceID is the name of a .wem containing a voice line of the initial category spoken by the specific character assigned by CAKSwitchCntr (at least from what I've seen thus far, I'm not quite sure how this accounts for all of the Tav voice options)
I recognize this is a lot, so hopefully I or someone else can figure out a way to automate some of this process
Thank you so much for this!!! I'll check more into it and add it to the guide (with your credit!).
Where can I find the file
Shared\Public\Shared\Content\Assets\Sounds\Vocals\[PAK]_Vocals\merged.lsf
so that I can extract it? I'm not finding it in the game folders
Edit: Figured out that you get the file path from the Shared.pak file and extracting that. Now I'm trying to figure out the next part with the ulActionID
Even without it, how did you convert the bnk to XML? Would you consider uploading the converted file as a modder's resource?
Trying to use wwiser for it and I just can't figure it out
Edit: I am going through the SharedSounds Sounds folder and it seems like it would be those .wem files in there.
Aren’t we lucky?! Zombieman4 has transcribed with AI, proofread over 4000 lines and shared them with us! Have fun!!
https://docs.google.com/spreadsheets/u/0/d/1Ba17_2G-sq7wcbXVLf66Lb_hQVBbPmf5/htmlview
My most heartfelt thanks for all of the research and work everyone in this chat has done - this is an extension of it and I couldn't have made the tool without it.
https://www.nexusmods.com/baldursgate3/mods/16462
826541967.wem, just not pan out no matter what you try? The original
has stereo audio at 44100Hz. I cut the new audio to match in
Audacity, export as .wav, convert in Wwise to Vorbis with matching
settings as the source, and no luck. It cuts out the audio in game,
but does not replace it, though the exported file can be played in
foobar2000 no problem.
I use this file as an example because there is an existing mod using it, and I thought,
if this mod works, it’s not cutting out because I can’t use audio
replacers. If I use the original mod, the sound is replaced just
fine, but even if I take their .wem, convert it to .wav, then back to
.wem, it no longer plays in game.
I have been going through trial and error for days trying to figure this out. I’ve
posted to the BG3Modding Reddit as well as the Discord, but no one
seems to know what is going wrong. Here's a screenshot of my
settings in Wwise when converting.
I’ve tried WEM Opus just for giggles. I’ve tried using different sample rates on
purpose, using versions 2022, 2023, and 2024 of Wwise, installing
Wwise on a different machine and converting there. All no luck. I
must be over looking something. I’ve followed the Replacing in-game Sound instructions from the Modding
Community Wiki, read Custom Battle Music Audio Replacement on Nexus,
and watched the Audio and Voice Replacer Guide. I just don’t know what else to do. I’ve done model edits and it’s so easy, I
must be missing something simple!
My durge voice 4 has been effectively exorcized of the new patch7 maniacal laughters and spoilery lines thanks to his wisdom and kind attention!
While still being a durge voice, with its occasional outbursts of creepiness, it does no longer sound deranged and out of control, just the way it
was pre-patch 7.
This was achieved by introducing blank sound files as replacements for the following:
525773031.wem & 601217730.wem = maniacal laughs
185320340.wem = father, they will die for you
274409118.wem = father, are you watching
134440127.wem = they are dying for me. all of them (this one was not spoilery but the high pitch on the "me" sounded so much like deranged Orin it just got on my nerves).
Files can be created as .txt files and just plainly change the extension to .wem
They should be placed in your Data\Public\Shared\Assets\Sound folder.
The result should be the aforementioned voicelines not playing any more.
THANK YOU SO MUCH AGAIN!