Releases: I5UCC/VRCTextboxSTT
Releases · I5UCC/VRCTextboxSTT
TextboxSTT v0.8.1
Download GPU Version (Faster, Higher Requirements, Bigger Performance Compromises)
Download CPU Only Version (Slower, Lower Requirements, Smaller Performance Compromises)
Requirements
With default settings, this program has following requirements:
- CPU version:
- ~2GB of storage space.
- ~1GB of available RAM.
- GPU version:
- CUDA enabled GPU (NVIDIA ONLY), otherwise it will fall back to using CPU.
- ~5GB of storage space.
- ~1GB of available RAM.
- ~1GB of available VRAM.
- SteamVR (IF ran in VR, no Oculus support as of now.)
v0.8.1 Changelog
- Fixed breaking issue that prevented TextboxSTT from running with VRChat simultaneously
- Fixed issue where TextboxSTT had to be run before SteamVR.
v0.8 Changelog
Most notable changes
- Added a SteamVR overlay!
- See your transcribed messages right in front of your vision!
- Turned off by default, make sure you turn it on in the settings, if you want to use it!
- Added different transcription modes!
- once (Default). How the program always used to function. Speak once and after you stop talking, transcribe once.
- once_continuous. Speak and continuously update as you are speaking until you stop talking.
- realtime. Always listen and continously update as you are speaking.
- Added energy threshold determination. press the ⟳ button next to the energy threshold textbox in the settings UI to automatically determine the energy threshold for your microphone!
- Replaced "Banned Words" with "Word Replacements". you can now replace any word with another of your choosing:
- Emote Support over KAT!
- Support for up to 80 Emotes!
- 15 emotes by default, rest sacrifices Japanese Hiragana for emote slots. Just replace the KAT texture with the replacement texture sheet provided in the programs files "KAT_Emote_Texture_Sheet" folder. Here are some examples:
- Added a "Reset OSC Config" button to the main Settings UI, to quickly reset VRChats OSC config. (Requires a restart of VRChat to take effect)
- #1 Added Translation from any language to English.
Fixes
- Fixed delayed messages when using mode 0.
- Fixed deletion of messages when using in game keyboard.
- Fixed auto detection of language.
Other changes
- transcribing now takes account of the last audio window when needed.
- moved all OSC related code to the OscHandler class.
- other refactoring of code.
TextboxSTT v0.8
Download GPU Version (Faster, Higher Requirements, Bigger Performance Compromises)
Download CPU Only Version (Slower, Lower Requirements, Smaller Performance Compromises)
Requirements
With default settings, this program has following requirements:
- CPU version:
- ~2GB of storage space.
- ~1GB of available RAM.
- GPU version:
- CUDA enabled GPU (NVIDIA ONLY), otherwise it will fall back to using CPU.
- ~5GB of storage space.
- ~1GB of available RAM.
- ~1GB of available VRAM.
- SteamVR (IF ran in VR, no Oculus support as of now.)
Note:
- With this release, starting this program without having SteamVR running will cause the program to go into a "Desktop only" mode. Make sure to start the program after SteamVR is running.
v0.8 Changelog
Most notable changes
- Added a SteamVR overlay!
- See your transcribed messages right in front of your vision!
- Turned off by default, make sure you turn it on in the settings, if you want to use it!
- Added different transcription modes!
- once (Default). How the program always used to function. Speak once and after you stop talking, transcribe once.
- once_continuous. Speak and continuously update as you are speaking until you stop talking.
- realtime. Always listen and continously update as you are speaking.
- Added energy threshold determination. press the ⟳ button next to the energy threshold textbox in the settings UI to automatically determine the energy threshold for your microphone!
- Replaced "Banned Words" with "Word Replacements". you can now replace any word with another of your choosing:
- Emote Support over KAT!
- Support for up to 80 Emotes!
- 15 emotes by default, rest sacrifices Japanese Hiragana for emote slots. Just replace the KAT texture with the replacement texture sheet provided in the programs files "KAT_Emote_Texture_Sheet" folder. Here are some examples:
- Added a "Reset OSC Config" button to the main Settings UI, to quickly reset VRChats OSC config. (Requires a restart of VRChat to take effect)
- Added Translation from any language to English.
Fixes
- Fixed delayed messages when using mode 0.
- Fixed deletion of messages when using in game keyboard.
- Fixed auto detection of language.
Other changes
- transcribing now takes account of the last audio window when needed.
- moved all OSC related code to the OscHandler class.
- other refactoring of code.
TextboxSTT v0.7 Alpha
Download GPU Version (Faster, Higher Requirements, Bigger Performance Compromises)
Download CPU Only Version (Slower, Lower Requirements, Smaller Performance Compromises)
v0.7 Alpha Changelog
Most notable changes
- Added different transcription modes!
- once (Default). How the program always used to function. Speak once and after you stop talking, transcribe once.
- once_continuous. Speak and continuously update as you are speaking until you stop talking.
- realtime. Always listen and continously update as you are speaking.
- Added a energy threshold determination. press the ⟳ button next to the energy threshold textbox in the settings UI to automatically determine the energy threshold for your microphone!
- Replaced "Banned Words" with "Word Replacements". you can now replace any word with another of your choosing:
- Emote Support over KAT!
- Support for up to 80 Emotes!
- 15 emotes by default, rest sacrifices Japanese Hiragana for emote slots. Just replace the KAT texture with the replacement texture sheet provided in the programs files "KAT_Emote_Texture_Sheet" folder. Here are some examples:
- Added a "Reset OSC Config" button to the main Settings UI, to quickly reset VRChats OSC config. (Requires a restart of VRChat to take effect)
Other changes
- transcribing now takes account of the last audio window when needed.
- moved all OSC related code to the OscHandler class.
- other refactoring of code.
TextboxSTT v0.6
Download CPU Only Version (Slower, Lower Requirements, Smaller Performance Compromises)
Download GPU Version (Faster, Higher Requirements, Bigger Performance Compromises)
v0.6 Changelog
- A bunch of bug fixes after v0.5.
- some under the hood fixes and improvements.
- Integration with KillFrenzyAvatarText!
- A Settings UI, so you dont have to fiddle around with the config.json anymore:
- An Entrybox for Text to Text purposes:
VRCTextboxSTT v0.4
Download CPU Version (Slower, Lower Requirements, Lower Performance Compromises)
Download GPU Version (Faster, Higher Requirements, Higher Performance Compromises)
Requirements
With default settings, this program has following requirements:
- CPU version (Slower, Lower Requirements, Lower Performance Compromises):
- ~2GB of storage space
- ~1GB of RAM
- ~2GB of storage space
- GPU version (Faster, Higher Requirements, Higher Performance Compromises):
- CUDA enabled GPU (NVIDIA ONLY)
- ~5GB of storage space
- ~1GB of RAM
- ~1GB of VRAM
- CUDA enabled GPU (NVIDIA ONLY)
v0.4 Changelog
- Now loads whisper models into the "whisper_cache" folder.
- Shows download progress of models in main text box.
- Async recording and transcribing to prevent the GUI from freezing.
- Added a "banned_words" key to config.json, add any amount of words that you want to filter out, case insensitive. ["word1", "word2", ...]
- Added "max_transcribe_time" to determine the maximum amount of time transcribing can take before canceling it.
- Code optimizations / refactoring.
TextboxSTT v0.3
Download GPU Version (Faster, Higher Requirements, Higher Performance Compromises)
Download CPU Version (Slower, Lower Requirements, Lower Performance Compromises)
v0.3 Changelog
- added audio feedback when clearing the textbox
- fixed another cancel feature bug (hopefully the last one)
TextboxSTT v0.2.3
Download GPU Version (Faster, Recommended)
Download CPU Version (Slower)
v0.2.3 Changelog
- Further refactoring
- added version to UI
- Fixed bug that didnt allow the program to run without a VR Headset connected or SteamVR installed.
TextboxSTT v0.2.2-Hotfix
Download GPU Version (Faster, Recommended)
Download CPU Version (Slower)
v0.2.2-Hofix Changelog
- Fixed issue where when a transcription is canceled, it instantly listens after.
v0.2.2 Changelog
- Added a "cancel" function, simply hold the button/hotkey in any action to cancel it.
- merged "record_hotkey" and "clear_hotkey" to one "hotkey" to act the same way as the controller button.
- UI launches before loading in whisper etc.
v0.2.1 Changelog
- refactoring and optimization.
- better logging.
- Fixed bug where the logs get appended to the log file forever.
- Fixed bug where UI initializes AFTER downloading the model. (now BEFORE it)
v0.2 Changelog
- Added a GUI!
- Added more Audio Feedback
- Added two options to the config:
-- "hold_time"
-- "timeout_time"
TextboxSTT v0.2.2
Download GPU Version
Download CPU Version
v0.2.2 Changelog
- Added a "cancel" function, simply hold the button/hotkey in any action to cancel it.
- merged "record_hotkey" and "clear_hotkey" to one "hotkey" to act the same way as the controller button.
- UI launches before loading in whisper etc.
v0.2.1 Changelog
- refactoring and optimization.
- better logging.
- Fixed bug where the logs get appended to the log file forever.
- Fixed bug where UI initializes AFTER downloading the model. (now BEFORE it)
v0.2 Changelog
- Added a GUI!
- Added more Audio Feedback
- Added two options to the config:
-- "hold_time"
-- "timeout_time"
TextboxSTT v0.2.1
v0.2.1 Changelog
- refactoring and optimization.
- better logging.
- Fixed bug where the logs get appended to the log file forever.
- Fixed bug where UI initializes AFTER downloading the model. (now BEFORE it)
v0.2 Changelog
- Added a GUI!
- Added more Audio Feedback
- Added two options to the config:
-- "hold_time"
-- "timeout_time"