FirstResponse GPT is a plugin for the Grand Theft Auto V modification LSPDFR that creates a dynamic police simulation environment using Large Language Models (LLM), Text-to-Speech (TTS), and Speech-to-Text (STT) services.
- Live Dispatcher Audio: Dynamically generated and voiced dispatch responses.
- Dynamic NPC Interactions: Utilizes LLMs to generate realistic and context-aware NPC dialogues.
- Voice Input Recognition: Integrates STT services to process officer voice commands.
- Realistic Dispatcher Responses: Employs TTS to produce lifelike dispatcher communications.
- Callout Awareness: Automatically generates dispatcher speech when new callouts are triggered.
- Hooking Support: Seamless integration with UltimateBackup for use with its additional backup types.
- Immersive Radio Effects: Applies audio filters to simulate authentic police radio transmissions using the companion JUCE framework plugin built for FirstResponseGPT.
This plugin is designed for content creators, immersive players, and LEO roleplay communities who want their in-game dispatcher to behave and sound more like a real person.
For example:
- You say: "Control, I'm on scene."
- The plugin sends your speech to the LLM.
- The LLM responds with something like:
Copy that, 10:22 hours.voiced through ElevenLabs with a radio-filtered effect. Or - You say: "Start me another unit! Fight in progress!"
- The plugin sends your speech to the LLM.
- LLM determines that not only are you requesting backup, but that this is a priority request.
- The LLM initiates a priority signal sound effect over the radio, responds with something like:
All units, officer requesting assistance. Any available officer response.voiced through ElevenLabs with a radio-filtered effect. - Other officers that the LLM generated over the radio begin responding to the dispatchers traffic, indicating they are responding and a backup request is triggered in game for each officer.
This is currently still a WIP and in development. Installation instructions are not provided at this time.
This project is licensed under the MIT License. See the LICENSE.txt file for details.
Note: This project is in active development. Contributions and feedback are welcome!