Neural Recognition

Choose how DRE listens: local recognition on your PC or cloud recognition when you want another option.

• Run local recognition with GPU acceleration when available for faster responses.
• Fall back automatically when local recognition is not available.
• Pick the model that fits your setup: faster recognition, higher accuracy, or a balance of both.

The goal is simple: reliable voice control that matches your PC, your connection, and your racing needs.

Use desktop speakers without DRE hearing itself, and interrupt it while it speaks. Acoustic Echo Cancellation removes DRE's own speaker output from microphone input. It is enabled by default under Sound > Input, while Push-To-Talk keeps the original microphone audio. Wired or USB microphones with desktop speakers are supported; Bluetooth, virtual audio cables and far-field setups are best effort.

Pause mid-command without DRE treating it as finished. Smart Voice Command Pauses can distinguish a completed command from an incomplete pause and keep listening while you finish. DRE still responds promptly when your command is complete, but gives you up to two seconds to continue after an incomplete pause, or up to five seconds while you are busy on track. This works with Neural or Hybrid Recognition, Voice Activated or Hotword Activated input, and VAD set to Both. If unavailable, DRE safely falls back to normal voice detection.

DRE prepares voice recognition during startup. In Hybrid mode, Neural can become available before Classic finishes loading, so voice commands work as soon as Neural is ready. Classic joins later, and the ready alert plays once. Local Neural can prepare in the background while cloud recognition remains available. If Classic is unavailable, Hybrid continues with Neural alone.