Faq
FAQ
Does this require a paid API?
There's a free tier of 2000 transcriptions per day. Most servers stay free indefinitely. Beyond that it's $0.04 per audio hour, which works out to about $5 a month for a 32 player server running 4 hours a day.
Will players need to pay for anything?
No. Players just press a key and talk. Everything happens server-side.
What about player privacy?
No audio is ever stored on disk by Stratis Voice. The client captures audio, sends it to your server, your server forwards it to the speech provider, the provider returns text. The audio is discarded after transcription.
The speech provider's privacy terms are linked in your purchase email.
Will it work on QBCore / ESX / vRP / standalone?
Yes. Stratis Voice is framework-agnostic. It hooks player-loaded events for QBCore, ESX, and falls back to a network session check for everything else.
Will it work on RedM?
Probably, but it's not officially tested. The audio capture and HTTP layers are the same in RedM.
Can players bypass the keybinds and constantly transcribe?
The server enforces:
- Per-player rate limit (12/min default)
- Minimum gap between requests (750ms default)
- Per-player daily cap
- Server-wide daily cap
- Audio fingerprint deduplication
Even if a player modifies the client to spam, the server rejects everything past the limits.
How accurate is it really?
Whisper Large V3 Turbo is the same model OpenAI ships. On clear English speech it's near-perfect. On RP slang, callsigns, and 10-codes, the accuracy depends on whether you've set up prompt biasing. With a good prompt, it transcribes "1-Adam-10 to dispatch 10-4" correctly the first time, every time.
How fast is it?
About 300ms from end of speech to action firing. Audio capture is realtime, the API call takes ~250ms, the rest is network.
What languages are supported?
50+ including English, Spanish, French, German, Italian, Portuguese, Polish, Dutch, Japanese, Korean, Russian, Mandarin, Arabic. Set Config.STT.language to any ISO 639-1 code, or nil for auto-detect.
Can I use it for admin tools?
Yes. Set minimumConfidence = 0.8 on admin actions and use a wake word like "stratis admin kick" so they only fire on deliberate input.
Can I add commands without restarting?
Yes. Use the NewAction export from any other resource. See API reference.
Does it conflict with FiveM voice chat?
No. Voice mode uses a separate keybind (default F7 + N). Your normal voice chat key is untouched.
What happens if the speech provider goes down?
The script returns a "service busy" notification and the action does not fire. Players just press the key again. The state machine has a 12-second watchdog so it never gets stuck.
Can I sell content built with Stratis Voice?
Yes. The resource is licensed for use on your servers and in resources you distribute, with the standard reseller restrictions in your purchase agreement.