Skip to content
View in the app

A better way to browse. Learn more.

LCPDFR.com

A full-screen app on your home screen with push notifications, badges and more.

To install this app on iOS and iPadOS
  1. Tap the Share icon in Safari
  2. Scroll the menu and tap Add to Home Screen.
  3. Tap Add in the top-right corner.
To install this app on Android
  1. Tap the 3-dot menu (⋮) in the top-right corner of the browser.
  2. Tap Add to Home screen or Install app.
  3. Confirm by tapping Install.

2 Screenshots

GrammarPoliceBridge — Voice-Control GrammarPolice in ANY Language

Fix GrammarPolice's language and speech recognition limitations.

GrammarPoliceBridge is a companion plugin for GrammarPolice (LSPDFR) designed to solve one of GrammarPolice's biggest limitations: language support.

Instead of relying on GrammarPolice's built-in speech recognition, GrammarPoliceBridge acts as a voice bridge between your microphone and GrammarPolice.

Simply hold your Push-to-Talk key, speak naturally in your preferred language, and GrammarPoliceBridge will recognize your speech, match it to a GrammarPolice command, and trigger the corresponding action automatically.

Speak Thai, English, German, Russian, Japanese, or other supported languages — and let GrammarPolice handle the rest.

HOW IT WORKS

Hold the Push-to-Talk key while speaking.

GrammarPoliceBridge listens through your microphone, converts your speech into text using your selected speech engine, matches the recognized text against the available GrammarPolice commands, and triggers the appropriate action.

The default PTT key is Right Alt, but you can configure it to use another key.

HOW TO USE

Step 1 - Your push-to-talk key
Open plugins\LSPDFR\GrammarPoliceBridge.ini. By default you hold Right Alt and speak, then release to send - the same key GrammarPolice already uses for dispatch. Any key works:

  [General]
  PTTKey=RMenu        // Right Alt. F9, F10, LControlKey also work
  PTTMode=Hold        // Hold = radio style, Toggle = tap once to start, tap again to send

Step 2 - Choose your engine

  [General]
  Engine=Vosk         // recommended: instant, very light on the CPU, works offline
  Engine=Whisper      // local AI, most accurate, heavier (needs a small download)
  Engine=Windows      // Windows built-in speech, needs the Windows language pack

Step 3 - Choose your language
Language= tells the bridge two things: which GrammarPolice grammar pack to read (from GrammarPolice\grammar\<Language>) and which speech model to use for you.

These are the packs that exist, so Language= must be one of them:
  de-DE, en-CALIFORNIA, en-GENERIC, en-IMMERSIVE, en-MET, es-ES, fr-FR, it-IT, ru-RU, th-TH

  English -> Language=en-GENERIC
  German  -> Language=de-DE
  Thai    -> Language=th-TH

Step 4 - Model quality (Vosk only)

  [Vosk]
  ModelQuality=accurate   // default: the most accurate model Vosk has for the language, 70 MB to 2 GB, downloaded once
  ModelQuality=small      // lightweight (~40 MB) for old PCs and slow internet

Step 5 - First run
The model is downloaded once, with a progress notification on screen. If your connection drops, the download continues where it stopped the next time you start the game instead of starting over.

Step 6 - Check that it worked
RagePluginHook.log states which model was loaded and which mode is active:

  Vosk model loaded: ...\GrammarPoliceBridge\Models\vosk\<model>
  Keyword phrases loaded: 160 (th-TH)

and the on-screen message "Ready - <n> phrases, <n> keywords" appears when the plugin is listening for you.

Step 7 - Change what a word does
Every command phrase is in GrammarPoliceBridge\keywords\keywords_<lang>.ini. Open the file for your language, add or remove a phrase under the action you want, save, and restart the game. Your callsign and GrammarPolice settings stay in GrammarPolice\custom.ini.

Example - playing in Japanese
Japanese is not one of GrammarPolice's own languages, but the bridge supports it. Use an English grammar pack for the actions and point the bridge at the Japanese speech model:

  [General]
  Language=en-GENERIC
  Engine=Vosk

  [Vosk]
  ModelName=vosk-model-ja-0.22            // 1.0 GB, most accurate
  ModelName=vosk-model-small-ja-0.22      // or the ~40 MB lightweight one

Then speak Japanese, for example 応援 (backup), コード3 (code 3), 逮捕 (arrest). Japanese command phrases ship with the plugin in GrammarPoliceBridge\keywords\keywords_ja.ini - the same is true for Thai, English, French, German, Spanish, Italian, Russian, Portuguese, Dutch, Polish, Turkish, Vietnamese, Indonesian, Korean, Chinese, Arabic, Hindi and Swedish. Commands in English keep working at the same time, so you can mix languages.

ModelName= is needed whenever the language you speak is not a GrammarPolice pack (Japanese, Chinese, Korean, Indonesian, ...), because the Vosk model follows Language= by default and you would otherwise end up with the English model.

FEATURES

  • Works with ANY language — GrammarPolice grammar pack of your choice (th-TH pack bundled)

  • 3 speech engines: Windows (built-in), Vosk (live recognition, light on CPU), Whisper (local AI, most accurate)

  • Keyword dictionaries for 19 languages with cross-language fallback — say "code 4", "โค้ดโฟร์" or "Code vier" and it understands

  • Hold-to-talk or toggle mode, configurable action delay for realistic dispatch response

  • On-screen "Listening..." feedback, optional recognized-text overlay

  • True single-file plugin — no Python, no servers, no extra DLLs

  • Speech models auto-download on first run (one-time)

QUICK START

  1. Install LSPDFR and GrammarPolice

  2. Extract the zip and copy the contents over your GTA V folder

  3. Edit GrammarPoliceBridge.ini: set Engine=Vosk (recommended, works offline, great for Thai)

  4. Launch via RAGEPluginHook — hold the PTT key and speak

REQUIREMENTS

CONFIGURATION

All settings live in plugins\LSPDFR\GrammarPoliceBridge.ini — language, engine, PTT key and mode, action delay, notifications.

TROUBLESHOOTING

  • No response? Check Windows microphone privacy settings (Settings > Privacy > Microphone)

  • First launch downloads the speech model — be patient (40–190 MB depending on engine)

  • Make sure GrammarPolice's custom.ini has EnableSpeechRecognition=false and DispatchKey=None (the included deploy script does this automatically)

CREDITS

Companion plugin for GrammarPolice.

Edited by KongUser

What's New in Version 1.0.3

Released

Fixes the engine failing to start with "LoadLibrary failed for libgcc_s_seh-1.dll (error 126)" on a clean install.

1.0.3 loads the engine files dependency-first with LOAD_WITH_ALTERED_SEARCH_PATH, so

Windows looks in their own folder where those files are. Checked on a PC with an empty

PATH: the old loading order failed all four files, the new one loads all four. No manual

deleting, no configuration.

The rest of 1.0.3 is the diagnostics that were missing:

- If the files are complete but Windows still will not start them, the message now names

the exact component it could not find instead of telling you to delete anything. If it

names MSVCP140.dll, VCRUNTIME140.dll or VCOMP140.DLL, install the Microsoft Visual C++

2015-2022 Redistributable (x64) and start the game again.

- The engine libraries' own import tables are read, so the log names the missing

dependency rather than a bare Win32 error code, and every loaded library is logged with

its folder and exact size.

- The Whisper engine loads its own native files now too. whisper.dll needs the ggml dlls

from the same folder, and Whisper.net's loader has the same limitation, so Whisper could

have failed the same way.

- The Windows engine now reports what it hears: the recording device in use, the speech

languages Windows has installed, whether your Language= is one of them, every

recognition result with its status and confidence, and an explicit "no text recognized

for this press" line. Before this, switching to Windows speech and getting nothing gave

you an empty log with no way to tell whether the microphone, the language pack, the

privacy setting or the engine was at fault.


Short Description

Fixes GrammarPolice's language and speech recognition limitations by bridging your microphone with GrammarPolice.

AI disclosure

The author has disclosed that AI tools were used in creating the following aspects of this file:

  • Description
  • Images
  • Mod and Mod Content (Moderate)

The plugin code was developed with significant AI assistance, and the cover image was AI-generated for the mod page. AI was meaningfully used in code, translation and documentation, though the design, testing and integration were done by the author.

User Feedback

Recommended Comments

gtrey

Members

Hey I keep getting an error vosk model download failed it then says that my source and destination path must have identical roots. what does that mean and how do I fix it?

KongUser

Members Author

6 hours ago, gtrey said:

Hey I keep getting an error vosk model download failed it then says that my source and destination path must have identical roots. what does that mean and how do I fix it?

Thanks for reporting this! I’ll look into it and fix it in the next update.

JMMA2506

Members

(edited)

hello I get a vosk download error and it does not allow me to use gramar

Edited by JMMA2506

gtrey

Members

1 hour ago, JMMA2506 said:

hello I get a vosk download error and it does not allow me to use gramar

Is this the error you are getting?

Could contain: text, screenshot, outdoor

JMMA2506

Members

No

KongUser

Members Author

(edited)

14 hours ago, JMMA2506 said:

hello I get a vosk download error and it does not allow me to use gramar


Can you attach your RagePluginHook.log here (or just paste the lines with "Vosk" in them)? That tells me exactly why the download is failing. Also let me know what your Language= is in GrammarPoliceBridge.ini.

Edited by KongUser

boski009

Members

1 hour ago, KongUser said:


Czy możesz załączyć tutaj swój plik RagePluginHook.log (lub po prostu wkleić linijki z „Vosk”)? To wyjaśni mi dokładnie, dlaczego pobieranie się nie udaje. Daj mi też znać, jaki jest Twój Language= w pliku GrammarPoliceBridge.ini.

I had a similar issue: over 50% downloaded, then an error occurred. The radio was working, though—when I pressed the call button, I heard a sound, but no one answered. So, I applied the settings you listed in the TROUBLESHOOTING section, but now "Grammar Police" isn't working; it says "radio disabled." Here is my log—can you help?

RagePluginHook.log

KongUser

Members Author

(edited)

6 hours ago, boski009 said:

I had a similar issue: over 50% downloaded, then an error occurred. The radio was working, though—when I pressed the call button, I heard a sound, but no one answered. So, I applied the settings you listed in the TROUBLESHOOTING section, but now "Grammar Police" isn't working; it says "radio disabled." Here is my log—can you help?

RagePluginHook.log

Please update to v1.0.2 and try again - that build fixes exactly this. The line that matters is "Vosk model failed to load: LoadLibrary failed for libgcc_s_seh-1.dll (error 126)": an engine file the plugin unpacks was left half-written, or was blocked by antivirus, and 1.0.2 now checks those files and rebuilds them itself instead of failing on every launch. No manual deleting needed.

Just extract the new zip over your GTA V folder (your ini, Models folder and custom.ini stay as they are), start the game and hold your PTT key - the log should then read "Vosk model loaded: ...".

And "radio disabled" is normal, not something the settings broke - that is GrammarPolice's own speech switched off, which is exactly how the bridge is meant to run.

Edited by KongUser

JMMA2506

Members

Hello I just downloaded the new mod but now every time I start I get this message and I already deleted the folder like 3 times

Could contain: text, man, person, clothing, display device, computer, monitor, outdoor

Ir0nH3art

Members

(edited)

1 hour ago, JMMA2506 said:

Hello I just downloaded the new mod but now every time I start I get this message and I already deleted the folder like 3 times

Could contain: text, man, person, clothing, display device, computer, monitor, outdoor

Im having the exact same issue with vosk. Get that long error message and then when i talk theres no radio sound and logs say no speech...

So i decided to switch engines from vosk to windows. I now get a promp saying its ready to use just press PTT and itll work.

I talk into the mic and get absolutely nothing. No speech back, no radio sounds and nothing in the logs saying its even attempting to read or talk back

Edited by Ir0nH3art

KongUser

Members Author

On 9/17/2026 at 12:09 AM, JMMA2506 said:

Hello I just downloaded the new mod but now every time I start I get this message and I already deleted the folder like 3 times

Could contain: text, man, person, clothing, display device, computer, monitor, outdoor

Sorry - that message is ours, not your install, and deleting that folder was never going to

help. It is fixed in 1.0.3, which I am putting up now.

What happened: one engine file (libgcc_s_seh-1.dll) needs another one (libwinpthread-1.dll)

sitting next to it, and Windows' loader does not look in that folder, so it fails on the very

first file with error 126. Nothing on your side causes it, and nothing you delete changes it.

If you want to play before 1.0.3 is up: copy libwinpthread-1.dll from

plugins\LSPDFR\GrammarPoliceBridge\natives\ into your GTA V folder (the one with GTA5.exe) and

start the game - that is all it needs. 1.0.3 does it for you.

On 9/17/2026 at 1:10 AM, Ir0nH3art said:

Im having the exact same issue with vosk. Get that long error message and then when i talk theres no radio sound and logs say no speech...

So i decided to switch engines from vosk to windows. I now get a promp saying its ready to use just press PTT and itll work.

I talk into the mic and get absolutely nothing. No speech back, no radio sounds and nothing in the logs saying its even attempting to read or talk back

The Windows engine has its own rules, and it can start, listen and hear nothing. Three things

have to be true: a microphone the game can open, the Windows speech language for your

Language= actually installed, and "Online speech recognition" switched on in Settings >

Privacy > Speech - dictation uses that, and with it off you get silence and no error.

1.0.3 (posting now) fixes the Vosk error, and it also makes this engine talk: the log now shows

your recording device, the speech languages Windows has installed, whether your Language= is

one of them, every recognition result, and a clear "no text recognized for this press" line.

Fastest way to be playing: update to 1.0.3 and keep Engine=Vosk.

cortezyaa

Members

great work. it's will be cooler if you add player animations and sound fx from grammar police

Create an account or sign in to comment

Account

Navigation

Search

Search

Configure browser push notifications

Chrome (Android)
  1. Tap the lock icon next to the address bar.
  2. Tap Permissions → Notifications.
  3. Adjust your preference.
Chrome (Desktop)
  1. Click the padlock icon in the address bar.
  2. Select Site settings.
  3. Find Notifications and adjust your preference.