Quickstart Guide

Installation

Download the ZIP file for HOSCY
Unpack it and start the executable
If asked to, install .NET Runtime (Please use the DESKTOP x64 version, it will not work otherwise)
Allow firewall and make sure your antivirus is not complaining
Turn on OSC in your radial menu in VRChat and rejoin the instance
HOSCY is ready to go, just select your preferred speech recognition mode in the "speech" tab and press the button labeled "Stopped" to start recognition

I highly recommend immediately switching from the "Windows Recognizer" to one of the following as they're much better

Whisper is a highly precise AI recognizer that uses both CPU and GPU for speech recognition

Download an AI model here
- If you have a powerful PC I recommend using Medium EN or Medium for multi-language needs
- For weaker PCs I recommend Base EN or Base
- Alternatively you could also go for a middle ground with Small EN or Small
- Worst case scenario you can also use Tiny EN or Tiny
Unzip the files
Go to the "Speech" Page and press the button "Edit list" next to the "AI Model" dropdown
Add the path of the .bin file, close the window and select the model in the dropdown
That's it. Starting these usually takes a while so make sure you set the correct microphone

Vosk is a quite precise AI recognizer that uses CPU for speech recognition

Download an AI model here
- If you have 5/6 GB of RAM you don't need I recommend going for vosk-model-en-us-0.22
- Otherwise use the weaker version with way less RAM usage vosk-model-en-us-0.22-lgraph
- Alternatively you could also go for a middle ground with vosk-model-en-us-daanzu-20200905
Unzip the files
Go to the "Speech" Page and press the button "Edit list" next to the "AI Model" dropdown
Add the path of the folder containing all files, close the window and select the model in the dropdown
That's it. Starting these usually takes a while so make sure you set the correct microphone

HOSCY has multiple "pages" with different settings and usages

Main contains all your information and allows you to quickly mute, clear and stop recognition
Input features a manual input box with configurable presets
Speech contains all settings related to speech recognition including shortcuts and replacements, as well as the recognizer picker
API is all about external services like translation or remote speech recognition
Output contains all settings about the textbox and TTS
OSC lets you control your OSC parameters and routing
Config mostly includes logging information

Shortcuts and Replacements on the "Speech" page let you replace words with other words and let you trigger commands. Replacement looks for certain words and replaces them with others and Shortcuts does the same but replaces the entire message instead
Voice Commands allow you to do things like clearing your textbox with "clear" or control your currently playing media with "media skip", "media pause", "media resume" and more
Media Display display what you are currently listening to above your head (on output page)
AFK Timer and Counters Display how long you've been AFK and count parameters
OSC Commands let you control OSC parameters using for example your voice
OSC Parameters let you control HOSCY via your avatars radial