Fallout4 Elevenlabs Auto Voice Generator

[Wolfe6900]

Fallout4 Elevenlabs Auto Voice Generator

View File

---

Overview:

Creating voice files can be a tedious and long process for the regular modder, so I made a tool that lets you connect to Elevenlabs, assign a voice to your voice type in the CK and mass generate the voice files.

What it does:

- Deciphers and filters dialogue from Fallout4's CK exports

- Connects with Elevenlabs by using your custom API key (requires a 'Starter' or 'Creator' paid version). 

- Auto Detects VoiceType from your Fallout4 mod and allows you to link them with an ElevenLabs voice.

- Allows you to select dialects, i.e., 'Irish', to replace words and make authentic dialogue.

- Allows you to select different voices for different emotions.

- Allows you to skip duplicate files

- Generates mp3 (fake .wavs) files for download from Elevenlabs

- Built in Converter to make those files readable for 44.1 kHz Fallout4 .wav files

- Allows you to generate .lip files

 

How to use (Generic):

1. Obtain an API Key from ElevenLabs.
2. Obtain additional requirements: FaceFXWrapper, ffmpeg, ffprobe, FonixData
3. Configure your paths
4. Assign your FO4 Voices to Elevenlabs voices
5. Generate voice samples
6. Convert those samples to .wav
7. Generate .lip files
8. Move to your appropriate mod folder and you're done

 

How to use (Advanced):

1. Extract .ba2 files from Fallout4 - Voice.ba2

2. Obtain an API Key from ElevenLabs.
3. Obtain additional requirements: FaceFXWrapper, ffmpeg, ffprobe, FonixData
4. Configure your paths
5. Fetch voice & emotion to clone
6. Convert .fuz to .wav
7. Create clone in Elevenlabs
8. Select a dialect
9. Assign your FO4 Voices to Elevenlabs emotion/voice
10. Generate voice samples
11. Convert those samples to .wav
12. Generate .lip files
13. Move to your appropriate mod folder and you're done

 

How to use (Tutorial):
*Unpack the .zip to see a tutorial on how to change the Player's voice with Cait's. 
[Warning:] This would cost upwards of 461,820 credits!

 

Links to external tools needed:

FaceFXWrapper: https://github.com/Nukem9/FaceFXWrapper/releases
ffmpeg: https://github.com/BtbN/FFmpeg-Builds/releases
ffprobe: https://github.com/BtbN/FFmpeg-Builds/releases
FonixData: {yourFallout}\Data\Sound\voice\Processing
Yakitori Audio Converter: https://www.nexusmods.com/skyrimspecialedition/mods/17765

 

Special Thanks to: 54yeg (for the .lip converter tool allowing me to use as a template/reference for generating .lip files.)

---

• Submitter
  Wolfe6900
• Submitted
  01/25/26
• Category
  Modders Resources
• Requirements
  Fallout 4, Fallout 4 Creation Kit

 

[vaultbait]

Not a question directly about this mod, but related... Does Elevenlabs manage to generate decently believable super mutant voices yet? Asking for a friend. 

[jlab7788]

Jesus how did you make this thing. If it will work properly, it makes me very happy.

[Wolfe6900]

16 hours ago, vaultbait said:

Not a question directly about this mod, but related... Does Elevenlabs manage to generate decently believable super mutant voices yet? Asking for a friend. 

If you were to take the Fallout 4 - Voices.ba2 and extract it with Archive2, then convert the .fuz to .wav, upload the super mutant .wavs into Elevenlabs and clone the voice, I'd image it'd do a pretty good job.

[Wolfe6900]

10 hours ago, jlab7788 said:

Jesus how did you make this thing. If it will work properly, it makes me very happy.

A lot of making smaller .pys with ChatGPT & Grok and slowly building more onto it. Right now, I forgot to upload ffmpeg with it so it may have problems converting. Currently working on a better version that can do what this does, allow you to add dialect (i.e., Irish may say 'arse' instead of 'ass'), and also generate the .lip files all in the same program. Also putting in a part where it reads the dialogueExport and classifies the emotion so you can make different 'voices' and assign a different voice per emotion. 

 

Hopefully I'll be able to get it out tonight or tomorrow.

[jlab7788]

5 minutes ago, Wolfe6900 said:

ChatGPT와 Grok을 이용해 작은 .py 파일들을 만들고, 점차 기능을 확장해 나가고 있습니다. 지금은 ffmpeg 파일을 함께 업로드하는 것을 잊어서 변환에 문제가 있을 수 있습니다. 현재 이 프로그램과 동일한 기능을 수행하면서 방언 추가(예: 아일랜드어에서 'ass' 대신 'arse' 사용) 및 .lip 파일 생성까지 모두 하나의 프로그램에서 처리할 수 있는 개선된 버전을 개발 중입니다. 또한 dialogueExport 파일을 읽어 감정을 분류하여 각 감정에 맞는 다양한 '목소리'를 만들 수 있도록 하는 기능도 추가할 계획입니다. 

 

오늘 밤이나 내일쯤에는 보내드릴 수 있을 것 같습니다.

Happy to see that. And if i can stop to make a voice in the middle with a button, then it's more useful then now. Thank you for your work.

[JB.]

20 hours ago, vaultbait said:

Not a question directly about this mod, but related... Does Elevenlabs manage to generate decently believable super mutant voices yet? Asking for a friend. 

I haven't used Elevenlabs to create any mutant voices because xVASynth is pretty satisfying. Just a quick note: there are two voicetype for super mutants to download. CrSuperMutant is the best. CrSuperMutant02 never worked well. So, my solution in my mod was to give all the supermutants the CrSuperMutant voicetype. 

 

But since your mod has a lot of dialogue with super mutants, you might prefer to do it with Elevenlabs, so as to clone the three voicetypes.

[JB.]

4 hours ago, Wolfe6900 said:

A lot of making smaller .pys with ChatGPT & Grok and slowly building more onto it. Right now, I forgot to upload ffmpeg with it so it may have problems converting. Currently working on a better version that can do what this does, allow you to add dialect (i.e., Irish may say 'arse' instead of 'ass'), and also generate the .lip files all in the same program. Also putting in a part where it reads the dialogueExport and classifies the emotion so you can make different 'voices' and assign a different voice per emotion. 

 

Hopefully I'll be able to get it out tonight or tomorrow.

Wow, that "all-in-one" tool is interesting. Thanks for the tool; I haven't been able to use it yet, but I will soon. I use one that also generates lip files, in addition to WAV files from Elevenlabs. It's called "DBVO VoicePack Maker". But unfortunately, it's made for Skyrim, so I have to create the lip files with another program (and then merge the WAV and lip files with yet another program... ugh).

[Wolfe6900]

3 hours ago, JB. said:

Wow, that "all-in-one" tool is interesting. Thanks for the tool; I haven't been able to use it yet, but I will soon. I use one that also generates lip files, in addition to WAV files from Elevenlabs. It's called "DBVO VoicePack Maker". But unfortunately, it's made for Skyrim, so I have to create the lip files with another program (and then merge the WAV and lip files with yet another program... ugh).

I finally finished up the new tool, and even made it look pretty with some vault-boy images.

 

I'll need to do some testing with it though to make sure it does everything I want it to do. In the trials (when I used 7+ individual .pys), I was able to change the Player's voice for Cait's voice. That took a lot of bouncing around folder's, plus a lot of credits from Elevenlabs etc. 

 

So, with this one, I tried to consolidate everything and give you one tool to do it all. Right now, it seems like it's stable and functioning, but I'll need a little time messing with it to find it's true capabilities and limitations. 

 

I've also been working on a "how-to" to accompany it. It's a massive tool to give you the power to essentially swap the player voice so it's been a project.

 

 

[Rob S Williams]

First, let me say if you follow the instructions closely, it just works. Kudos Wolfe. It's still going to take some effort to dial in your workflow depending on the exact use-case, but this makes dialogue generation WAY more efficient. I was using F5-TTS and the results don't hold a candle to Elevenlabs. With the former the sound quality was trash and it'd take dozens of generations and a lot of splicing together of clips to get even semi-normal sounding speech. With latter, between 1-5 generations and you've got dialogue that almost passes for vanilla. Elevenlabs isn't expensive and is definitely worth the money.

 

The only issue I've run into is the accuracy of the .lip files and I'm wondering if I'm doing something wrong. I previously tried LipFuzer, which I'm guessing is what this tool is using, and the mouth movements are a full half second off at best and completely off from the actual spoken dialogue at worst. Same with the .lip files from this tool. The only way I've found to get accurate .lip files is using the 32-bit version of the CK, which is a complete nightmare. Am I the only one with this issue? The dialogueExport I'm using is the one that contains my custom NPC's dialogue and the text matches audio in the .wav files so I'm not sure what else would cause this.

 

Any thoughts?

 

 

 

 

[Wolfe6900]

On 2/3/2026 at 2:01 AM, Rob S Williams said:

First, let me say if you follow the instructions closely, it just works. Kudos Wolfe. It's still going to take some effort to dial in your workflow depending on the exact use-case, but this makes dialogue generation WAY more efficient. I was using F5-TTS and the results don't hold a candle to Elevenlabs. With the former the sound quality was trash and it'd take dozens of generations and a lot of splicing together of clips to get even semi-normal sounding speech. With latter, between 1-5 generations and you've got dialogue that almost passes for vanilla. Elevenlabs isn't expensive and is definitely worth the money.

 

The only issue I've run into is the accuracy of the .lip files and I'm wondering if I'm doing something wrong. I previously tried LipFuzer, which I'm guessing is what this tool is using, and the mouth movements are a full half second off at best and completely off from the actual spoken dialogue at worst. Same with the .lip files from this tool. The only way I've found to get accurate .lip files is using the 32-bit version of the CK, which is a complete nightmare. Am I the only one with this issue? The dialogueExport I'm using is the one that contains my custom NPC's dialogue and the text matches audio in the .wav files so I'm not sure what else would cause this.

 

Any thoughts?

 

 

 

 

I've noticed the .lip files came out a little strange, too. Not too sure how they're processed differently, but it's still a lot easier than manually going through and doing each file. I think a lot of modders just entirely skip voice files since it's very time-consuming so I'm hoping we'll see more mods with some voice files. 

[JB.]

I've already tested it. It's a wonderful tool. It saves me SO MUCH time. 🙏🥰

 

 

 

I only have two questions.

1- Can I change the program's color? I can barely see the letters.

 

Spoiler

 

2- What type of Elevenlabs voice does it use by default? EnglishV2, Multilingual, Flash? 

Edited March 11 by JB.

[JB.]

@Wolfe6900 I think I've found a problem. It's not generating the dialogue when it's more than one line. Lines two and three weren't generated, but they are in the txt file.

 

 

 

 

this is the txt file

 

 

 

 

 

 

I checked my Elevenlabs history and it didn't generate those two lines.

 

those dialogues are always marked as _2, _3, etc. 

 

 

Edited March 11 by JB.

[Wolfe6900]

On 3/11/2026 at 10:44 PM, JB. said:

@Wolfe6900 I think I've found a problem. It's not generating the dialogue when it's more than one line. Lines two and three weren't generated, but they are in the txt file.

 

 

 

 

this is the txt file

 

 

 

 

 

 

I checked my Elevenlabs history and it didn't generate those two lines.

 

those dialogues are always marked as _2, _3, etc. 

 

 

That would make sense as to why it seems to stop randomly. I'll have to look into how it deciphers that garbled garbage the CK pushes it. I think the tool just throws it out if it can't figure out how to decipher it, and it was pretty difficult to get it to figure out the layout. I've recently started moving away from using ChatGPT and focus more on Cursor for projects like this, so I may see if I can get it to figure out more in depth of how that .txt file is laid out.

[Dojo_dude]

I've been super busy making fuel [don't worry I won't get blown up 😆], but I just want to say amazing idea to create this tool! The Super Mutant voices finally being unlocked allows so much more potential for stuff like Approach, Violate, ect 👍