For years, you’ve had to speak DRE’s language. Now, DRE 2025.1 speaks yours. I’ve overhauled the voice recognition system, so you can forget memorizing commands and just focus on your driving. Let’s talk about what’s new.

For the longest time, talking to DRE felt a bit like casting a magic spell. You had to say the exact magic words, in the exact right order, or… nothing. “Auto fuel now!” worked, but “Hey DRE, figure out my fuel for the rest of the race” would just get you confused silence. Frankly, that was getting old. My goal is for you to be immersed in the race, not fumbling through a mental look-up table of commands.

That’s why I’m thrilled to introduce a massive overhaul to DRE’s brain: Neural and Hybrid Speech Recognition.

Speak Your Mind!

Forget the rigid command list. DRE now understands what you mean, not just what you say.

This new system doesn’t just match sounds to a list of phrases. Instead, it transcribes everything you say and then uses a custom-trained language model, which I’ve dubbed Thyrring, to figure out your intent. It’s built to understand racing-related prompts in all 7 supported languages. So now, you can say things like:

• “I need fuel to go to the end”
• “Calculate the fuel automatically”
• “What’s the fuel situation?”

And DRE will know you want it to run the auto fuel calculation. This has shown to work for over 80% of any DRE or racing-related prompts I’ve thrown at it. It’s not magic, but it’s pretty darn close.

Your Rig, Your Choice: Three Engines for Your Voice

You now have full control over how DRE listens to you. In the settings, you’ll find a new option to choose your speech engine, letting you balance performance and accuracy based on your hardware.

• Classic: The good(?) old Windows Speech Recognition. It’s fast, lightweight, and perfect if you’re a creature of habit who has all the commands memorized.
• Neural: This is the new hotness. It processes your speech locally on your own PC, using your GPU or CPU. It’s incredibly accurate and what enables the new intent-based commands. You can even choose from different models to balance speed and quality, from “Fast & Alright” for older rigs to “Slow & Great” for beefy GPUs.
• Hybrid: The best of both worlds. This mode uses the Classic engine to quickly check for an exact command match. If it doesn’t find one, it hands your speech over to the powerful Neural engine to figure out what you meant. It’s the recommended setting for most racers.

There’s also a Cloud option for Neural Recognition. This offloads the heavy lifting to remote servers, which is a fantastic choice if you’re on an older PC or want to save every last frame in VR.

The Little Things

This update isn’t just about the big stuff. I’ve added some quality-of-life improvements too. There’s a new subtle audio signal so you can hear when DRE is processing your speech. And if DRE ever mishears you, the “Report bad speech recognition” feature in Chat view now includes a handy play button so you can hear what DRE heard, making it easier to correct the transcription and help improve the system.

While DRE is much smarter, it still has its limits. It’s guard-railed to focus on racing, so asking it for a lasagna recipe will probably get you a polite refusal. It also can’t handle multiple requests in one sentence yet, so you’ll need to break complex commands into smaller pieces. And no, it still can’t cook your eggs… but it’s cooking! This update marks a huge leap in how you interact with DRE. It’s smarter, more flexible, and more natural. No more memorizing—just racing and talking. Ready to give your vocal cords a workout? Download the latest DRE version!