DoubleCheeseburger's xVASynth Based Mod Voicification Thread

[LostDremoraDream]

On 6/19/2021 at 4:54 PM, DoubleCheeseburger said:

How to Make Voice Files Yourself

 

This is my approach to making voice files for various mods written step by step.

During this guide I'm gonna generate voice files for Death Alternative - Your Money or Your Life for Skyrim Special Edition which comes with daymoyl.esm as an example, but the same process applies for any other plugin. Moreover, it should work Skyrim Legendary Edition and probably Fallout 4 too.

 

 

 

Getting Ready

 

Through out this guide I'm gonna be using various software.
For your convenience here is the complete list of tools I'm gonna be using:

• xVASynth for text to speech synthesis. More specifically, SKVA Synth;
• FFmpeg for post processing of generated audio files;
• xEdit (also known as SSEEdit) for exporting dialogue lines from the plugin;
• Mod Organizer 2 for mod management (but any mod manager should do);
• Creation Kit x64 with SSE CreationKit Fixes installed and FonixData.cdf from Legendary Edition placed in "Skyrim Special Edition\Data\Sound\Voice\Processing" directory for lip sync (.lip) files generation;
• Unfuzer CPP Edition for packing .wav and .lip files into .fuz;
• Sublime Text for configuring dialogue export script (but any text editor will do too).

There are plenty of guides on the Internet on how to install every program from this list, so I'm not gonna cover this here.

 

 

 

Getting xVASynth Ready

 

It's important to configure xVASynth so it generates voice files in the appropriate for Skyrim (and probably Fallout 4) format.

 

1. Launch xVASynth.

 

2. Open settings by clicking on the cog icon on top right:

  Hide contents

 

3. Make sure that the "Use ffmpeg post-processing (requires ffmpeg >=4.3)" checkbox is checked.

Make sure that the "Audio format (wav, mp3, etc)" field contains "wav".

Make sure that the "Audio sample rate (Hz)" field contains "22050".

Make sure that the "Audio bit depth" field contains "pcm_s16le":

  Hide contents

 

 

 

Getting xVASynth Dialogue Export xEdit Script Ready

 

This is my personal xEdit script that allows to easily export dialogue lines from any plugin in the xVASynth ready format.

It also offers easy way to update voice files whenever plugin or xVASynth voice list get updated without necessity of doing the whole process from scratch by creating cache files.

 

 

Installation

 

1. Download archive from this post.

2. Unpack it into your xEdit folder next to other scripts. In my case it's "F:\Program Files (x86)\SSEEdit\Edit Scripts".

 

 

Configuration

 

Before you can use the script there is some configuring that needs to be done.

 

1. Open xVASynth Dialogue Export.pas with any text editor. The Initialize function contains some basic script settings.

2. Find skvaVoiceIds.CommaText. This property basically contains comma separated list of available voice models in the <xVASynth Voice ID>=<Game Voice ID> format. Add or remove voices from the list according to your xVASynth voice models list.

 

3. Find maleVoices.CommaText and femaleVoices.CommaText. These properties contain the list of comma separated game voices grouped by sex. Add or remove voices from the list according to your xVASynth voice models list.

4. Optionally find perFileLimit. This script is able to split finalized dialogue lines into multiple files, because xVASynth really struggles with huge files. On my PC xVASynth can handle 50000 lines per file pretty stably. You might want different lines limit so change perFileLimit accordingly.

5. Optionally find filterText. If True, the script will filter out some parts of the text using the regular expressions listed below. Generally, you want to keep this variable True.

6. Optionally find cachePath. Adjust path for the cache files.

7. Optionally find outputPath. Adjust output path for the xVASynth ready batch files.

 

 

 

Creating Voice Files

 

Exporting Dialogue Lines

 

1. Launch xEdit.

 

2. Load your load order.

 

3. In the plugin browser find the mod you're making voice files for. In my case it's daymoyl.esm.

 

4. Right click on it and choose "Apply Script...":

  Reveal hidden contents

 

5. In the Apply Script popup find "xVASynth Dialogue Export" script and press OK:

  Reveal hidden contents

 

6. Wait while script does its job. The more dialogue lines in plugin the longer process will be. The process will be slower if cache file for the plugin is present, because script will have to compare each dialogue line against those contained in cache file.

One the process is completed the script will print something like this in the "Messages" tab:

  Reveal hidden contents

The generated file(s) will be placed in the "xVASynth Dialogue Export" xEdit folder. In my case it's "F:\Program Files (x86)\SSEEdit\Edit Scripts\xVASynth Dialogue Export\daymoyl.esm-1.csv".

 

 

Synthesizing Voice Files

 

First, let's create output folder where we'll store synthesized voice files.

1. In Mod Organizer 2 click on the screwdriver and wrench icon and select "Create empty mod":

  Reveal hidden contents

2. In the appeared popup window enter mod title. It can be anything you like, in my case I've chosen: "Death Alternative - Your Money or Your Life 7.0.1a.2 Voice Files".

  Reveal hidden contents

Press "OK".

3. Find newly created mod in the mod list and open it in Explorer:

  Reveal hidden contents

4. Create "sound" folder and enter it.

5. Create "Voice" folder and enter it.

6. Create folder named after plugin name you're creating voice files for and enter it. Make sure to use full plugin name including extension. In my case it's "daymoyl.esm".

7. Copy your current explorer location. In my case it's "D:\MO2\Mods\Death Alternative - Your Money or Your Life 7.0.1a.2 Voice Files\sound\Voice\daymoyl.esm". It's gonna be output path for synthesized voice files.

 

Now it's time for speech synthesis using xVASynth itself.

 

1. Launch xVASynth.

2. Open batch synthesis by clicking on the table icon on top right:

  Reveal hidden contents

3. Paste previously copied output path. In my case it's "D:\MO2\Mods\Death Alternative - Your Money or Your Life 7.0.1a.2 Voice Files\sound\Voice\daymoyl.esm":

  Reveal hidden contents

4. Drag .csv file made by xEdit script and drop it into xVASynth's drop zone. xVASynth will need some time to process the file if it's big, so be patient.

5. Once .csv file is processed press "Synthesize Batch" and wait while xVASynth does its job.

  Reveal hidden contents

xVASynth will start generating .wav files.

 

• Tip: xVASynth really struggles with rendering content of huge files. So once you started synthesis feel free to close "Synthesize Batch" popup by clicking anywhere outside of it. This will speed things up. You can still track the progress via Windows taskbar.

 

 

Creating Lip Sync Files

 

This where things can get really frustrating due to buggy nature of Creation Kit. Thankfully, there is easy option available as well if you don't care about lip sync.

Moreover, you can skip this step altogether but missing lips animation will look really weird in game.

 

Option A: Using Creation Kit

 

This option is the best in terms of quality, but the hardest and the most annoying part of this guide.

 

1. Activate mod that contains voice files in Mod Organizer 2.

 

2. In Mod Organizer 2 configure new executable for Creation Kit. In the "Arguments" field type -GenerateLips:<Plugin name>. In my case it's "-GenerateLips:daymoyl.esm". My final executable settings looks like this:

  Reveal hidden contents

 

3. Launch new Creation Kit executable. Creation Kit will start to generate .lip files for the specified plugin. Generated files will be placed to the overwrite mod. In my case it "C:\Users\admin\AppData\Local\ModOrganizer\SkyrimSE\overwrite\sound\Voice".

Eventually, unless you're really lucky, Creation Kit will come across faulty file that it can't really process and crash. Restarting Creation Kit will result the same. The only solution I know is to copy successfully generated lip files to faulty voice and restart Creation Kit.

 

4. Eventually Creation Kit will go through all dialogue lines and close itself. Now it's time to verify the result.

First, compare number of files in xVASynth output directory with number of files in Creation Kit output directory. In my case it's  "D:\MO2\Mods\Death Alternative - Your Money or Your Life 7.0.1a.2 Voice Files\sound\Voice\daymoyl.esm" and "C:\Users\admin\AppData\Local\ModOrganizer\SkyrimSE\overwrite\sound\Voice\daymoyl.esm". Ideally, both numbers should be the same. But sometimes Creation Kit refuses to generate .lip files for some reason. The is no unified process of figuring out why this happens. In most cases it's due to faulty dialogue conditions. Some plugins modify dialogue lines from the original Skyrim (which is weird on itself considering that original voice line will remain untouched), in this case Creation Kit won't generate .lip file too.

 

5. Once everything is verified and resolved, copy .lip files from Creation Kit output directory ("C:\Users\admin\AppData\Local\ModOrganizer\SkyrimSE\overwrite\sound\Voice\daymoyl.esm" in my case) to xVASynth output directory ("D:\MO2\Mods\Death Alternative - Your Money or Your Life 7.0.1a.2 Voice Files\sound\Voice\daymoyl.esm") respecting the folder structure. If everything is done correctly every .wav file is be paired .lip file:

  Reveal hidden contents

 

I use bunch of Mod Organizer 2 and Windows CMD scripts to handle most of this automatically and verify the result. But I'm not ready to share it, because those scripts have their own problems and the whole process is just a mess.

I heard that the author of xVASynth is going to create .lip generation functionality inside of xVASynth. Let's hope that this will actually happen. Having such feature would make this process so much easier.

 

Option B: Using Fake .lip File

 

This option is the easiest, but lip sync will be terribly off. Moreover, there is a good chance of NPC lips still moving after the end of audio file which will look weird.

 

1. Download Stock_10.lip from this post and place it in xVASynth output directory ("D:\MO2\Mods\Death Alternative - Your Money or Your Life 7.0.1a.2 Voice Files\sound\Voice\daymoyl.esm" in my case).

 

2. Launch Windows CMD (on Windows 10 press Win + S and type "Command Line").

 

3. In the command prompt type:

cd /d "<xVASynth output directory>"

In my case it's:

cd /d "D:\MO2\Mods\Death Alternative - Your Money or Your Life 7.0.1a.2 Voice Files\sound\Voice\daymoyl.esm"

Press Enter.

 

4. In the command prompt type:

for /R %G in (*.wav) do copy Stock_10.lip "%~dG%~pG%~nG.lip"

Press Enter.

This command will copy Stock_10.lip over to every .wav file with appropriate file name.

 

5. Stock_10.lip can be deleted now.

 

 

Packing Files in .fuz

 

1. Launch Unfuzer.

 

2. Point it to the earlier created Mod Organizer 2 mod folder in the "Folder to parse" option. In my case it's "D:\MO2\Mods\Death Alternative - Your Money or Your Life 7.0.1a.2 Voice Files".

 

3. Make sure that "Parse subfolders", "Recursively" and "Process .lip files" checkboxes are all checked.

 

4. Press "Refuz (Wav->Xwm->Fuz)".

 

Unfuzer will start to pack .wav and .lip pairs into single .fuz file.

 

 

Finishing

 

All done! Your mod should be voiced now.

Previously generated .csv file(s) (in my case daymoyl.esm-1.csv) can be (and probably should be) deleted now. But keep cache file in the "Edit Scripts\xVASynth Dialogue Export\cache" folder. It will come handy when you decide to update voice files for new version of the mod or new xVASynth voices.

 

Optionally share your work with others. 

 

 

xVASynth Dialogue Export.zip 2.68 MB · 32 downloads

Stock_10.lip 9.35 kB · 17 downloads

xVASynth Dialogue Export v2 Experimental.zip 2.93 MB · 15 downloads

is there a way you can provide similar info for Fallout 4?

[variaNTR]

Edited: I think they all work fine, since Babo 4.02 was released some time ago and I am using it along with 3.901 voicepack.

 

 

I wonder if the voice packs are fully compatible with newer version of mods.

BaboDialogue and Sexlab Defeat by BakaFactory just got recently updated.

 

Babo 4.02 This was released some time ago so I guess the recently changed to the file in download section was about description.

Sexlab Defeat Baka Edition 1.11

Edited August 3, 2021 by variaNTR

[DoubleCheeseburger]

On 8/3/2021 at 5:08 PM, variaNTR said:

I wonder if the voice packs are fully compatible with newer version of mods.

 

It should be compatible, unless the mod author changed the dialogues drastically in the updated version. So the text just won't match with the audio.

 

I'm working on the updated version of the voice packs to include more SKVASynth voices and take mod updates into the account too. The process is rather slow, because I just don't have much of spare time these days, but it should be all done by the end of the next week.

 

It's also great to see more people doing stuff like this, thanks everyone!

[DoubleCheeseburger]

1. Removed Make Skyrim Straight Again from the first post as the mod author provides voice files with the mod itself now.

2. Updated the guide post. Now it features .wav files preprocessing using SynthUtil by @Executaball . Finally pain free .lip files generation. 

3. xVASynth Dialogue Export v2 script is no longer experimental as it seems to be working fine.

 

 

On 8/2/2021 at 3:45 AM, LostDremoraDream said:

is there a way you can provide similar info for Fallout 4?

Is there something specific you're having problem with? I haven't tried it with Fallout 4 mods, but I'm sure that the guide would apply well to it as it is. Not so sure about xEdit script though.

[Executaball]

31 minutes ago, DoubleCheeseburger said:

2. Updated the guide post. Now it features .wav files preprocessing using SynthUtil by @Executaball . Finally pain free .lip files generation. 

 

Thanks for the mention in the guide, I see you removed the original files after preprocessing though? Do note that the final resampling rate for the process is 16kHz instead of the normally used 22.05kHz / 44.1kHz. I would recommend against using the processed audio as actual final voice files as the quality is noticeably lower. I usually only use them to generate .lip files in CK, then use the original (unprocessed) files for final .fuz creation.

[Executaball]

Also been helping out for adding batch UI pages, optimized csv parsing, and multi-threaded FFMPEG output processing in the next major version of xVASynth. Processing time should be greatly improved (30%-40%) (there will also now be a time estimate). 

 

 

 

Tested with 300k+ files processed as normal with no more UI lag~

We're also trying to have built-in .lip and .fuz generation within xVASynth, since Nukem9 (author of SSE Creation Kit Fixes and FaceFXWrapper) was able to help with modifying the wrapper for use by third party programs. With any luck this should be able to be multi-threaded as well. (As the current Creation Kit lip gen could very well be the longest part of the synthesis process) (i.e. 300k .lip files in CK currently takes 1-2 days). Though this might take a little bit longer.

Edit - Processing speed improvement:

Edited August 14, 2021 by Executaball

[LostDremoraDream]

5 hours ago, DoubleCheeseburger said:

1. Removed Make Skyrim Straight Again from the first post as the mod author provides voice files with the mod itself now.

2. Updated the guide post. Now it features .wav files preprocessing using SynthUtil by @Executaball . Finally pain free .lip files generation. 

3. xVASynth Dialogue Export v2 script is no longer experimental as it seems to be working fine.

 

 

Is there something specific you're having problem with? I haven't tried it with Fallout 4 mods, but I'm sure that the guide would apply well to it as it is. Not so sure about xEdit script though.

Only thing I have had problems with is my own 50/50 hit or miss shot at generating appropriate lip files due to Fo4 CK being very hacky...  

I was thnking a script for Fo4Edit would help streamline the process because working with voice in the CK is so goofy

[Laakeri]

The SexLab Solutions Revisited Voice Files has some lines working with Ostim version of SL Solutions Revisited, but some doesn't work. For example, Brynjolf's voice doesn't work when you are approached by him to join Thieved Guild. Anyone got fix for this?

[DoubleCheeseburger]

23 hours ago, Executaball said:

 

Thanks for the mention in the guide, I see you removed the original files after preprocessing though? Do note that the final resampling rate for the process is 16kHz instead of the normally used 22.05kHz / 44.1kHz. I would recommend against using the processed audio as actual final voice files as the quality is noticeably lower. I usually only use them to generate .lip files in CK, then use the original (unprocessed) files for final .fuz creation.

Makes sense, edited the guide. Thanks for pointing it out.

 

22 hours ago, Executaball said:

Also been helping out for adding batch UI pages, optimized csv parsing, and multi-threaded FFMPEG output processing in the next major version of xVASynth. Processing time should be greatly improved (30%-40%) (there will also now be a time estimate). 

Tested with 300k+ files processed as normal with no more UI lag~

We're also trying to have built-in .lip and .fuz generation within xVASynth, since Nukem9 (author of SSE Creation Kit Fixes and FaceFXWrapper) was able to help with modifying the wrapper for use by third party programs. With any luck this should be able to be multi-threaded as well. (As the current Creation Kit lip gen could very well be the longest part of the synthesis process) (i.e. 300k .lip files in CK currently takes 1-2 days). Though this might take a little bit longer.

These are wonderful news!

 

17 hours ago, LostDremoraDream said:

Only thing I have had problems with is my own 50/50 hit or miss shot at generating appropriate lip files due to Fo4 CK being very hacky...  

I was thnking a script for Fo4Edit would help streamline the process because working with voice in the CK is so goofy

Well, my xEdit script can't help with lip files generation. For now we have to deal with Creation Kit to generate lip sync files.

 

15 hours ago, Laakeri said:

The SexLab Solutions Revisited Voice Files has some lines working with Ostim version of SL Solutions Revisited, but some doesn't work. For example, Brynjolf's voice doesn't work when you are approached by him to join Thieved Guild. Anyone got fix for this?

You can't really fix it for now as SKVASynth just doesn't have every voice model from Skyrim available. Brynjolf's voice is one of the missing ones.

[DoubleCheeseburger]

Finally finished updating voice packs.

1. Packs are now featuring all new DLC2MaleUniqueNeloth, MaleBandit, MaleDrunk, MaleEvenTonedAccented, MaleUniqueDelvinMallory voice models.
2. Some of the packs got significally smaller since I used the latest version of my xEdit script.
3. Prison Overhaul Patched no longer requires any edits to male jailer since MaleEvenTonedAccented is now available.
4. BaboDialogue is updated to version 4.02.
5. SexLab Approach Redux is updated to version 1.802.
6. Prison Overhaul Patched updated to 5.01.

If you have some of the packs already downloaded, here are the update archives.

• BaboDialogue: https://mega.nz/file/LjpVUIiT#f0u2u9AqAzTdT6BXSjmpyWbcb9Qk-CtvcykqDRCCTUU
• Death Alternative - Your Money or Your Life: https://mega.nz/file/SygjRSoY#SCqZ2X89KNGqF73Hzuw5OXIw09-_PSckperiShlXwvA
• Slave Girls by hydragorgon: https://mega.nz/file/Xn5nHQzL#L7Mh3gsxVEumt_C6P3MMKTys-XZGZsfif_Q-rCbrtzI
• Prison Overhaul Patched: https://mega.nz/file/G2phmAzR#Afz8EC31fRyWvwudz2bhjfAVjVYE-oh0cc66CqR8b7I
• Radiant Prostitution/Gigolo: https://mega.nz/file/zupAlJKZ#_RozcT7V7F-33HjcGxKiQq6r4wN8WQcNjGOgGnh1N1o
• SexLab Approach Redux: https://mega.nz/file/ezxWAIyY#YTlMMv-jmFq96TywIQ-6kf3Fl58YbyhSw7onXd5f7Z4
• SexLab Confabulation: https://mega.nz/file/y25CRL4a#U_8Mfxg80F7i9ZETIycYfUUfRmr6YdWjsQBdj_kroqM
• SexLab Defeat: https://mega.nz/file/jrxl3SDK#Vl-syaW5aTWeChPQXe5cAFyyEUlp4ZUwN2XUnZN8GqA
• SexLab Eager NPCs: https://mega.nz/file/Lq5CWZ6C#SjbMhCrIA6Ai7LXk_2kzVXy0wovwTjYhE3QdF975_Ok
• SexLab Horrible Harassment Expansion: https://mega.nz/file/OrgRRKAQ#8Kfe9q2vBBRTxI7YYJP6bUwEesSK07EkIaCUbWh4E0U
• SexLab Romance: https://mega.nz/file/u75EzKoK#KGeBJjdmpQr9oO587c73fH-Jyc9eg7uX39MXflEJYok
• SexLab Solutions Revisited: https://mega.nz/file/Gz4G3YZI#k_miMIDdOjsDutOtlBMYVVGIIwSlQqyqtlSIYdTfuTY
• SLSF Fame Comments: https://mega.nz/file/Xrg2VJKY#uKE75RHjHLV8CZwdqByP9hY0JPkOT79B290eGztsCps
• Spank That Ass: https://mega.nz/file/SnhTgYbb#2HlDMdZJr5VMzO7AlZ50qve_nSjdsX_5kS_q-AcmS5o
• SexLab TDF Prostitution and Pimping: https://mega.nz/file/26hyyBTZ#1T3ctEX-PNNKbmAUACCgq2mewT5P6y7K4hnj-uO6lrk

Links will be available for 30 days from now.

But consider redownloading full package from the main post as some of the packs got smaller and will take less disk space.

 

Have fun!

[GamerLat]

There is Voice Pack for M.C.G. (Extended Social Interactions) 1.11.09 Mod?

[Mister X]

@DoubleCheeseburger There's an updated version of MaleSoldier voice, too, but the download is faulty, state now (posting time)

 

Within the corresponding .json file, you need to change a bit:

emb_size from 0 to 2

emb_i from 2 to 0

modelVersion from 1.1 to 1.3

 

Source: https://discord.com/channels/794590496202293278/794594881863548948/876633735745929296

[qualcheduna]

The voice packs are amazing and I really wanted to thank everyone involved for their creation.

 

However I wanted to ask if the voice pack for Babo Dialogue can be available as two split files, because it is above the 5GB limit for non-premium Mega users.

 

Thank you again for the mod

 

[Oblivion_Cat]

On 8/23/2021 at 7:23 PM, qualcheduna said:

The voice packs are amazing and I really wanted to thank everyone involved for their creation.

 

However I wanted to ask if the voice pack for Babo Dialogue can be available as two split files, because it is above the 5GB limit for non-premium Mega users.

 

Thank you again for the mod

 

 

Just click on download file at the bottom of the pay for premium popup. It'll download anyway. I just did it and i don't have premium mega.

[qualcheduna]

18 hours ago, Oblivion_Cat said:

 

Just click on download file at the bottom of the pay for premium popup. It'll download anyway. I just did it and i don't have premium mega.

 

Wow, it worked! I did not expect that.

Thank you very very very much

[Temphyx]

@DoubleCheeseburger Great idea and great effort, thanks!  It's probably been mentioned, if so apologies, only issue that's stood out is the lines that use 'ok".  The software sees this as 'ock' rather than 'o-kay',  Line(s) using this come up frequently in SLEN, for example.  Seems like this may not be an easy fix, but you're the better judge of that.

Thanks again for a game changer!

[Herowynne]

1 hour ago, Temphyx said:

software sees this as 'ock' rather than 'o-kay'

 

I always imagine the NPC is speaking with a Scottish accent, "Och!"

[XarisZ]

Is anyone working on voices for Relationship Dialogue System?

[NNS10]

On 6/13/2021 at 5:46 PM, Executaball said:

(My own VASynth voice packs have been moved to its own post, link here)

BSA packed DoubleCheeseburger voice packs:

 

So I realized I was getting 5 minute launch times in MO2 with a couple of these installed, figured it was the file transfer speed with MO2's vfs.

I've packed most of the voice packs I'm using into BSAs which should reduce space and improve load times, for me went from 3-5 minute Skyrim launch time back to a few seconds.

- Any loose files like Sexlab Survival has should automatically overwrite since the BSA loads first. If you'd like to use the voices within the pack over the original voice files, you just need to delete the sound/voice folder within the original mod.

- Prison overhaul was the only mod that came with its own BSA, so I unpacked and merged it with the voice pack. The original BSA only had 2 texture files related to hair so I assume it will continue to work for the near future, just make sure there are no changes to POP's BSA file when installing new updated versions and it will continue to work.

- There is a notable size reduction but the obvious point of this is to reduce the file numbers. For example SL Survival's voice pack has 66,476 files and 47 folders. With Babo dialogue coming to 174,736 files.

 

**New conversions**

SL Survival DoubleCheeseburger BSA Voice Pack V0.662 beta
Loose file size (1.83 GB) reduced to BSA (1.72 GB), Archived (1.56 GB)

SE (ESL): https://bit.ly/3wswvdu

LE (BSA only, load with MO): https://bit.ly/36xPfxH

 

 

**These BSA below are packaged for SE only**

 

RadiantProstitution DoubleCheeseburger BSA Voice Pack v5.0
Loose file size (1.39 GB) reduced to BSA (1.32 GB), Archived (1.20 GB)
https://mega.nz/file/dBpEFB7Z#_4ryYw-oH7QkdR_PsFgwIj2_uvg2n4Vr6g4mF9OuiBQ

 

Spank That Ass DoubleCheeseburger BSA Voice Pack v4.5
Loose file size (242 MB) reduced to BSA (213 MB), Archived (182 MB)

https://mega.nz/file/YAhlSIwA#AVDN7wvDk0WPJ3zmVUid9181EzdYn5c4qK7sOvBir60

 

SL Defeat Baka Edition DoubleCheeseburger BSA Voice Pack v1.02
Loose file size (316 MB) reduced to BSA (285 MB), Archived (251 MB)
https://mega.nz/file/1IwAjBTY#_7V0Khd6ywNTn8a86w6G2phoYs0TtTIqzATQQP0F0fE

Prison Overhaul DoubleCheeseburger BSA Voice Pack V5.0
Loose file size (744 MB) reduced to BSA (674 MB), Archived (596 MB)
https://mega.nz/file/5BA0UD5S#f97OV3q6OycMwxzf0vqMzziHZ9TEcsDO_ALlxsdUESU

 

(Split) Babo Dialogue DoubleCheeseburger BSA Voice Pack (Full) V3.901
Loose file size (6.84 GB) reduced to 4x BSAs (6.52 GB), Archived (5.97 GB)

https://mega.nz/file/UZhWnYDA#XHLXgTaTloHMEabJMdHjf5B0_Rq9vOnVtapknZU_q8k

SLHH Expansion DoubleCheeseburger BSA Voice Pack V3.44
Loose file size (234 MB) reduced to BSA (219 MB), Archived (199 MB)
https://mega.nz/file/RcgFiK4R#f71whsCZOws1Ro4Xv4ojPayJmvAq1dt6alSk2YvQO7Y

SL Approach Redux DoubleCheeseburger BSA Voice Pack V1.70
Loose file size (1.55 GB) reduced to BSA (1.47 GB), Archived (1.34 GB)
https://mega.nz/file/Fd4CxRoa#yDRhAgq0gCAdX7mGuKowmkQEmr2VHrWZwaSPn1Tlb4M

 

These voice packs are really large. The vanilla game's voice BSA is 1.7GB while the voice packs for some of these mods are of similar or larger size.

 

How is that possible? Is VAsynth generating a lot of unused voices?

[Roggvir]

4 hours ago, NNS10 said:

These voice packs are really large. The vanilla game's voice BSA is 1.7GB while the voice packs for some of these mods are of similar or larger size.

How is that possible? Is VAsynth generating a lot of unused voices?

It can be big if the voices are uncompressed, but most probably it is as you think - there could be plenty of surplus voice lines generated, which are never going to be heard in game, because none of the NPCs using that voice can ever get into a situation where they would be eligible for saying the line.

 

For example, if some lines are supposed to be said only by a follower (ie. because only player's "follower" can become his/her master, etc., and some voice types are never used by any followers by default), then that may significanlty limit the voices available as not every NPC is/can be a follower, but MAYBE the lines for those voices got generated too.
If that is the case, it is no fault of the xVASynth - it generates whatever the user feeds it with.
But it is often hard to do the voice lines export while including only the voices that can ever use the lines, because almost none of the mods were made with that in mind, so the dialogues in them are lacking conditions that CK (or some xEdit script) can use to limit the pool of available voice types for that dialogue (or individual topics/branches/ or even just individual lines).
So, when you are doing the dialogue export to feed to xVASynth, unless you know a mod really really well so you can add those dialogue conditions yourself, you end up with a list of lines for voice types that may never speak those lines in the game.
 

That being said, it may not be the case here, or not with all those packages, i don't know what's in them.
Also, some mods like the Radiant Prostitution, may require almost all voice types anyway, because (and i only imagine, i never used that mod myself) who can become a client isn't probably limited enough (i imagine its basically anybody who isn't hostile), so with those kind of mods you cannot limit the export to just some voice types, because as a result of allowing virtually anybody to be a client, it may be needing almost all voice types anyway.

 

Furthermore, it may not be fair to compare the download sizes to the size of voice files in vanilla game.
In vanilla game, you have many dialogues that are limited to just a handful of NPCs, or in case of quests it is often just one NPC that can ever say certain lines.
In vanilla game, you have only one Camila Valerius who says certain lines while telling you how to get to Bleak Falls Barrow, but in modded game, you have about 40 different voices that can say "How much for a blowjob?" (i just made that up, but you see my point, right?)

Edited September 7, 2021 by Roggvir

[Mister X]

4 hours ago, NNS10 said:

 

These voice packs are really large. The vanilla game's voice BSA is 1.7GB while the voice packs for some of these mods are of similar or larger size.

 

How is that possible? Is VAsynth generating a lot of unused voices?

 

I assume, there are two reasons for that.
One, there really are quite a few unused voice files generated by xVA Synth. The script DoubleCheeseburger made to export the lines from the mods tries to find out which voicetype has this line. This isn't truly optimized I think, so for some lines where only a few voicetypes can use it, it's still generated for every voicetype xVA Synth has installed. Let's say, you have every possible voicetype installed for xVA Synth, but a line could only be spoken by 4 of them, you had over 80 unused voice lines in the worst outcome. Still, nobody wants to search for those manually ^^

 

Two, at least for Skyrim LE, and I think SE, too, the voices are very heavily compressed to save space, but this lowers the audible quality, too. Compare it to High Quality Voices, this provides THREE bas files for the vanilla game, just because the voices are not as compressed.

 

Ok, Roggvir beat me to it with a way more in-depth explanation ^^

Edited September 7, 2021 by Mister X

[Roggvir]

9 minutes ago, Mister X said:

Two, at least for Skyrim LE, and I think SE, too, the voices are very heavily compressed to save space, but this lowers the audible quality, too. Compare it to High Quality Voices, this provides THREE bas files for the vanilla game, just because the voices are not as compressed.

Well, i hope you are wrong about the last one.
I mean, who in their right mind would NOT compress the audio generated by xVASynth for fears of loosing quality?
Besides, even if somebody would be that much insane as to think it will lower the quality, they can still compress the files using higher quality settings.
In short, there is exactly ZERO reasons for not compressing the files produced by xVASynth.

[Mister X]

9 minutes ago, Roggvir said:

Well, i hope you are wrong about the last one.
I mean, who in their right mind would NOT compress the audio generated by xVASynth for fears of loosing quality?
Besides, even if somebody would be that much insane as to think it will lower the quality, they can still compress the files using higher quality settings.
In short, there is exactly ZERO reasons for not compressing the files produced by xVASynth.

 

You got me wrong. I don't say that people used UNcompressed voices, but they use settings that produce higher quality than the base game provides.

 

Simply the fact, that Bethesda used 48kbps for compression, whilst normally people set 192kbps or at least 128kbps, has a well audible impact. High Quality Voices just changes the vanilla voice lines from PC (48kbps) with those from XBox (192kbps). Why Beth used other compression settings for Xbox? No idea, but I'm thankful that those quality lines exist.

 

And that are the settings that normally are used in xVA Synth, too. 192 or at least 128kbps, instead of 48 of the base game files. So, OTHER compression, a bit LESS, but not NONE  

Edited September 7, 2021 by Mister X

[NNS10]

On 5/19/2021 at 5:37 PM, DoubleCheeseburger said:

Hi hello,

 

Recently, I've spent some time on generating voice files for my Skyrim build using xVASynth. The result proved to be a great immersion enhancer even though the quality of some of the synthesized speech is not that great.

This thread exists to share the result of my work with others.

 

Important Things to Understand

1. I don't tune lines to synthesize or edit generated audio files. This requires whole another level of commitment and time that I simply don't have. So if you use this, be ready to experience sound files that range from atrocious to nearly perfect to hilarious.
2. xVASynth is still lacking some of the generic Skyrim voice models. So don't expect 100% coverage from these voice files.
3. There are dialogue lines in some of the mods in which asterisks and brackets are used to accent some words in a sentence and in other lines of the very same mod it's used to describe NPC's actions, sounds, etc. For example:
   "The man that lives there is *very* friendly and you don't even have to pay."
   "*Before you have time to react he reaches down pinches a tuft of your pubic hair, pulls it out roughly and holds the hairs up to your face*"
   I decided to remove all text contained within asterisks, brackets, square brackets and "<...>" contructions from speech generation. Obviously, this will lead to some minor problems when certain words will be missing from audio even though logically they should be pronounced.

 

Thanks for the work on this.

 

Another thing to consider for #3 is also replacing "ok?" with "okay?" I heard a line with pronouned as auk. Though maybe this could also be considered a grammar issue for the mod author to fix.

[torn]

Will Fallout 4 sex mods be voiced?