VRChat 2019.2.5

Release - Build 825 - 31 July 2019



  • We've revamped and updated our audio system! Audio spatialization has been re-tuned and several new features and components are available
    • A focus on improvement of user voice, fixing issues with spatialization, and improving audio overall
    • Improved the default voice falloff parameters
    • Added a compressor/limiter to avatar audio channel to prevent maliciously loud audio
    • Added a (separate) compressor/limiter to user voice channel to prevent maliciously loud audio
    • Added a low-pass filter for voice audio at long range
    • Some adjustments to the Gain on various sound channels for better audio balance
    • Removed "Voice Prioritization" option
    • Changed voice falloff behavior. For details, see our documentation regarding our voice falloff curve
    • Check out the SDK section for changes relevant to world and avatar creators, namely the addition of the VRC_SpatialAudioSource and VRC_PlayerAudioOverride components
    • Further changes to prepare for later audio upgrades
  • Added some new summer emoji!


  • Streamlined audio mixer
  • There are now only 5 volume sliders: Master, Menu, Voice, World, and Avatar
    • Default values for these sliders will be adjusted on first launch:
      • Master: 100%
      • Menu: 100%
      • Voice: 75%
      • World: 70%
      • Avatar: 60%
  • Master volume setting can be overdriven to 125%. Going over 100% will cause the text to turn red. Be very careful setting Master above 100%!
    • This change is intended to accomodate for various hardware setups. If you swap devices, re-check this setting
  • Removed menu "cyber" background sound
  • Optimized use of UI audio sources for Quest
  • Added controller wireframe models for new Touch controllers, Index controllers
  • Improved IK performance in low-FPS situations
  • Added new World Report reasons
  • Updated to Oculus Unity Integration 1.37
  • Updated video helper binaries (three times, just to make sure)


  • Fixed an issue where blendshape visemes would sometimes get stuck locally after mic input was completed
    • This fix also involves a very small change in blendshape viseme behavior. Your default model state should be "non-speaking", and the sil blendshape should be essentially unchanged from that, aside from one vert slightly moved to prevent Unity from removing the empty blendshape on import. This is because we now reset all viseme blendshapes to zero when speaking is complete. We don't expect this to impact many users, but we're putting this info here just in case
    • As an aside, the sil blendshape is what is used when silence is detected but the mic is still open. When mic activity stops (the icon goes out), all viseme-related blendshapes are set to zero.
  • Fixed an issue preventing Fixed Foveated Rendering from working on Oculus Quest
  • Fixed an issue where Line Renderers were being counted improperly in the Avatar Perf Stats system
  • Fixed an issue where tooltips on controllers were sometimes incorrect
  • Fixed an issue where some components were not reliably blocked using the Safety system


  • Safety and Security fixes
  • Further iteration on Networked IK

Known Issues

  • If a GameObject on an avatar with a 2D Audio Source starts as disabled, it will be 3D when enabled. To work around this, enable/disable the Audio Source component, not the GameObject it is attached to
  • Some menu sounds may be slightly louder than others in some situations
  • Valve Index Controller users may find that their controllers are stuck in their hands during some launches
  • Quest users may encounter audio "slowing down" in particularly crowded instances



  • Small flow improvements to the account creation and verification process
  • Error page added for cases where an attempt to verify an email address fails
  • Improvements to the ToS agreement process on the VRChat website



  • Added VRC_ObjectSync.Respawn, which is accessible via the SendRPC event on VRC_ObjectSync
  • Allow multiple Triggers/Trigger types on the same VRC_Trigger component
  • VRC_Trigger will now default to "Default" layer mask and "TriggerIndividuals"
  • Audio Changes
    • Added component VRC_PlayerAudioOverride, to allow customizing voice and avatar audio behavior using Trigger zones or globally. You can use this to customize scene-wide falloff, create a performance stage for wider-reaching audio, or adjust falloff based on the part of the world the user is in. Check the documentation for details
    • Added component VRC_SpatialAudioSource which replaces ONSPAudioSource. Check the documentation for details
    • Using the "Enable 3D Spatialization" button in the Build Control Panel will now apply VRC_SpatialAudioSource instead of ONSPAudioSource (and will convert existing ONSP sources)
    • Removed "custom voice falloff range" from world descriptor, this option is no longer supported. Use a VRC_PlayerAudioOverride with Global toggled on instead.


  • Fixed a bug where onStation triggers weren't appearing in list
  • Fixed issue where sometimes importing the SDK would cause Unity Editor to hang

Did this page help you?