Skip to content

2.0.0 Beta 1

Pre-release
Pre-release

Choose a tag to compare

View all tags
@FallTalk FallTalk released this 11 Jul 14:46
· 58 commits to main since this release
2.0.0-Beta1
33d69ed
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.
  • Automatically Voice ESL/ESP files using the ESP Voice Generation. This uses a custom made xEdit script to find all the voice lines in a mod and automatically generate them for you. Makes creating vocied protagonist mods require 3 clicks.
  • Added support for new models: F5, CSM, LLASA, Spark, Orpheus, Chatterbox, FishSpeech, DIA, GPT SoVitts Super
  • Added AI audio Super Resolution upscaler, which makes the lower quality audio (16k, 24k) sound much better. The difference is noticeable. This is currently only used for generation, not the bulk enhancer.
  • Added support for mega models: CSM, Spark, and Orpheus have been trained on ALL of the voices in Fallout 4. These generate very good voices, and generally RVC is not needed so its been disabled at default.
  • Upgraded to CUDA 12.8 to support 50 series GPU
  • Show estimated VRAM each model uses
  • Make transcription optional, greatly increasing speeds. Transcription button removed and will be done automatically as needed.
  • References are now optional, if one is not selected a default will be used.
  • Reference bar color now reflects if enough reference length is selected.
  • All pages (Generation, References, Bulk, etc) now contain a help and settings flyout for easy use. This should help explain things for beginners and make changing settings much faster.
  • Bulk generation is now faster with logic improvements.
  • Downloading models will now give feedback in the load spinner showing how many files are needed
  • Added the ability to disable huggingface SSL, which should fix issues people have with being unable to download models
  • Fixed Error messages to make them more clear, I hope.
  • Upgraded to latest elevenlabs api
  • Upgraded to latest versions of lots of libraries.
  • Creates a new log file every time you start, for easier tracking.
  • Only saves the last 10 logs, to prevent it from taking tons of storage.

Removed:

VoiceCraft, MusicGen and SoundFX Gen as they added a lot of bloat. VoiceCraft audio editing now done by F5 TTS, which is much faster and better.

https://huggingface.co/falltalk/falltalk4/resolve/main/release/FallTalk_v2.0.0_beta1.7z

Assets 2
Loading