You can become a member or buy tokens from your account to enjoy Speech To Text as well as Speech To Text To Speech features from the program. The Text-to-Text feature is free in our program.

The VRCSTT program runs on Windows 7, 8, 8.1, 10 & 11.
You might also be trying to run the program from a drive with insufficient permissions. Try changing its location to a different drive. Also verify you are not missing a Windows library to run certain desktop program.

The AI voice response should normally take around a second to process before you can hear it. If the delay is longer, this is due to higher latency or connectivity issues that you may be encountering or a server encountering a temporary unexpected high load.

The server is able to handle around 60 message requests per second. Currently, we are using a little less than 1 message per second during peak weekend hours with an average turnaround of 100-105ms per request. This gives us around 0.5% failed requests in total.

Error message: Error loading discovery document: Error connection to https://dev-lmscvcu8.us.auth0.com/.well-known/openid-configuration. A task was canceled..

If you happened to have received the following, this is due to an old hash key that failed to reset properly after 30 days. You will need to perform the following:

1. In Windows Search, type regedit and Run as administrator
2. Navigate to: Computer\HKEY_CURRENT_USER\SOFTWARE\VRCSpeechToText
3. Locate the field CheckedSalt, right click and select “Modify…”
4. Empty the “Value data” field then press Ok.

Error message:

This error message will show when the .exe is unable to find the other .dlls or other files needed for the program to work properly.

Try and verify your drive for permission issues or try and run the program from another location.

If you are having an issue where it says the DLL is not found despite having the .exe in the correct location, it is highly possible that you may be missing a relevant C++ Redistributable that needs to be installed on your computer. The following can be downloaded here: https://docs.microsoft.com/en-us/cpp/windows/latest-supported-vc-redist?view=msvc-170

You will need to relogin inside the program for it to detect your new tier. You can do so by going inside Settings and clicking on Re-login. Once done, you should be able to use the STTS tab afterwards.

Latency could be the cause of your issue. If you speak too quickly after pressing the record button, the translation might miss the first word you say.

After pressing the record button, you can try leaving a small 300ms or so before saying your first word. The same would also apply to the end of your sentence if you feel that your sentences are cut too quickly.

The STT Program .NET core warning is something that happens if your Windows software doesn’t have the .NET 3.1 library installed on your computer. This library is a Microsoft Windows Framework that enables you to run existing web/server applications or in this case existing Windows desktop applications.

To fix this issue, click on “Yes” when prompt to download the library or visit the following link: .NET Core 3.1.

Click on the download button for the “Run desktop apps” that applies to your windows software.

Once downloaded, click on the .exe and run through the installation process.

And there you go, you should now be able to run the STT Program without issues!

If for some reasons you receive this warning:

This means that your computer is unable to run the program from the specific location it is currently installed. Try and move the extracted content from the .zip package in another location (example: Downloads, Documents, Desktop).

We use Cloud-Based services. We buy commitment packages from them to reduce costs per generated messages. This way ensures you can enjoy flawless and swift speech recognition services at a very low cost.

SteamVR must be running. SteamVR makes the connections between your Oculus and the VRCSTT program when checking for your bindings. If you’ve recently modified your bindings, try closing/reopening the STT program.

No!!!!!!!! >:(

No, you do not need a discord account. If you only need Text-to-Text you can use the program for free. If you need it for Speech services you can either buy tokens from the website or subscribe to our Patreon.

The Speech Bubble needs a bare minimum of 43 bytes and goes up in increments of 8 for faster bandwidth. You can also enable/disable the following options:
Manual resizing: 8 bytes
Language support: 7 bytes

Verify and make sure you have the latest version of the VRCSDK. Older versions of the SDK will not work with the Unity package. If after importing you still do not see it, make sure you have no errors in your console.

To hide the “Show My Avatar“, try scaling down the Game object: “!STT/World Constraint Container/Local Constraint Container/Speech Bubble” to 0,0,1.

The blend files are only used for custom edits purposes. They are not needed for the plugin to work on your avatar.

If you receive an error while trying to import the unity package in your assets folder due to the blend files, uncheck them from the list before importing the package.

If you do not have enough parameter space for the STT plugin you will have to delete a few parameters that you can part ways with.

A lot of avatars also use ints for toggles rather than using bools. If you have a basic understanding of Unity, you can convert those ints into bools and apply the change in your Animator layers and Animator parameters.

When applying the STT plugin for the first time on your avatar, a Reset Gestures layer will be created. This layer will contain an animation that specifically tells your avatar what the default blendshape/expressions on your avatar are supposed to be.

This process is automated and in almost all cases, you do not need to change anything to it. Depending on the avatar though, it is possible that you might need to go and manually add some stuff inside the animation layer. Example: wings, ears, hidden blades, etc…

Prior to making any changes when you click on the “Update Speech To Text” component, the component verifies your Animator layers and will look for layers that contains the name “Gesture“. Once found, it will verify if any of your gestures contains “Write defaults” that are checked.

If it finds one, a warning will appear asking if you wish to create a “Reset Gesture” layer to fix the Write defaults problem that your avatar is currently facing. Press “Yes” to create that layer.

Once done, you should have a new layer created in your Animator that will have a single motion/animation containing all your avatar’s default blendshapes.

Certain avatars have issues with some body parts not allowing you to attach objects to it. By default, the script adds an empty game object called “!STT Avatar Position Constraint” that is located within the “Armature/Hips/Spine/Chest“.

Instead, try dragging it onto “Armature/Hips/Spine/Chest/Neck/Head” if you are facing this issue.

Then once this is done, apply the position constraint.

Verify your controllers located in the Bindings tab of the STT program. If you have rebinded certain keybinds in VRChat, make sure they also reflect the same changes in the program.

For example, if you rebinded your “B” button to “A” and want to use your “A” button for speaking, you will need to use the “B” button instead.

Also, make sure the right microphone is used in the STT Program.

The script has to be removed from the avatar after setting up the Speech to Text component. You can remove it manually or look at the error that says ”The following component types are found on the Avatar and will be removed by the client: SpeechToText” and click AutoFix.

The program requires stable servers and stable clients. If the VRC servers are having issues, or if the person has less than 10 FPS or spotty connection, they will experience issues reading your messages. As this problem is related to server or personal connection issues, there is no solution on our side, and you may have to repeat messages on occasion.

It is also highly possible that your Messagebox Update rate is too high. The default update rate is set to “400“. If needed, you can increase it to “500” or more and see if this solution works for you.

Your antivirus may cause the program to freeze the first time it’s ran due to the antivirus scanning it. This is a normal behavior. You can wait a few seconds or close the STT program in Task Manager and start it again. This has been observed mostly with Avast antivirus.

Making an exception in your virus scanner for the program is not needed.

Make sure that both Avatar/Character bandwiths are the same in the Speech To Text component (Script) and also the STT program.

If they are the same, try reimporting the STT.unitypackage in Unity and reupdating your avatar. To make sure the Unity package is reimported properly, make sure you have deleted the old package (_STT) located under your Assets folder:

If you happen to have done some edits to your Speech Bubble and you have encountered some weird small white lines, verify your texture.

That’s a mipmapping issue. Check the text texture currently used on your Speech Bubble and copy those settings. Specifically mipmapping, and compression.

You can find the texture image at “Assets/_STT/Textures“. The normal settings should be the following:

Hope this helps!

There can be multiple reasons for this behaviour. If your avatar happened to have write defaults on your toggles, you will need to follow this documentation to fix your toggles and other related animations: https://docs.google.com/document/d/1LBzv1jzLN7PlTPxDH1KtQ7TnJZ8O_s76F6yQESkpf00/

If you didn’t receive a message, there is a possibility that your toggles/transforms are outside of the FX layer, which will not be taken in charge when the STT applies its automated fix on all FX animations/motions. (Gesture layer = shapekey modification only)

If you have transform manipulation or toggles on any controller but the FX controller, you need to copy those layers onto the FX controller. If you  have shapekey stuff AND transform stuff in the animations, just move those onto the FX controller as well.

Everything should work up to this point, although if you want to follow the VRC documentation, it is preferable to split those animations and apply them separately. The copied layers should go above the STT stuff. Most of the time it won’t need to, but it’s better to be safe than sorry.

When the speech bubble is populating with text, the parameter “STTGroupToggle” will be anywhere between 1 and 240. Once the message is done populating, it will go idle at 255.

In Unity, create a new animation layer in your Animator (example: “Play Bubble Sound”). Inside, create a new state that will contain a new animation inside.

Inside the new animation, set active an audio component/sound that will play if “STTGroupToggle” is between 1 and 240. If it is 0 or 255, set the audio component to disable.

If done correctly, it’ll make noise when the speech bubble is writing/populating then go back to being idle when it has completed.

Try switching your STTS service for something else than Azure. Microsoft Azure and similar products under Microsoft have approached their end of the line on Windows 7. There is not much we can do about it other than ask you to either upgrade your current Windows version or to use another STTS service from the choices we offer.

VRC Fury has a feature called “Fix Write Defaults“.

This feature breaks the Speech bubble and can cause other issues. To make sure it doesn’t touch WD, the Fix mode needs to be set to Disabled.

If you have already uploaded your avatar once with the wrong Fix mode, detach the current Blueprint ID from your avatar and reupload your avatar as a new blueprint.

If you are using an avatar that has the STT component installed, your VRCSTT Program will work with it.

This way, if you happen to have a friend that does Avatar work/commissions, you can share the STT.unitypackage with that creator and they can install it on an avatar that you will be using.

Small reminder that the STT & STTS are features where you need to be subscribed to RabidCrab’s patreon or be a VRCSTT user with tokens under your account for the STT Bubble to display what you say. If not, you can always use the Text-to-Text feature for free.

1. Open CMD as administrator and type the following command: netstat -aon | findstr "9000"
2. Locate the line “UDP <LocalIP>:9000 …” and look at the end of the line for the process ID
3. Open Task Manager and go into Details. Look for that PID.
4. If you’ve found the Process ID but it isn’t VRChat terminate that process. It is haggling the port and we cannot use it.
5. If you cannot find the process ID, it was shut down incorrectly and you need to perform a PC reboot.

This behaviour is due to the fact that the STT program is unable to pick up your voice. Verify and make sure that the correct microphone is selected in the “STT” tab.

This bug could be because of many different reasons.

– Check your console for any errors/warning that you might have with shaders (ex: Poyomi);

– If you have used the Pumkin tool, make sure that the copied Speech Bubble’s root bones are used on the newest avatar and not the old one;

– Reset OSC while in VRChat

– Try changing worlds (sometimes certain worlds can interfere with the speech bubble. It is very rare but could be the issue you’re facing.)

If all fails after this point, delete the “_STT” folder found inside your Assets, reimport the latest STT.unitypackage and try updating the Speech To Text component on your avatar a second or third time.