2025.1.2.3

🔄 Text-To-Speech

• Shaved off 200-600ms delay in TTS cloud cache retrievals. Enjoy the even faster responses now.
• Added verbose logging for certain users that face long delays before cloud cache reports back.
• More graceful logic handling of missing files in the cloud cache.

✅ Prevented duplicate update installer launches

• Fixed an issue where the app could accidentally start the same update installer multiple times.
• If auto-install is enabled, DRE will now properly detect and avoid re-launching an update installer that is already running.

✅ Neural Recognition

• Issue with loading CUDA and Vulkan libraries once downloaded the first time
• Optimized performance in general by core recognizer updates and bug fixes
• Extended support for Nvidia GPUs using CUDA for faster neural recognition and better handling when
  • Not having an Nvidia GPU
  • CUDA required files don’t download or place themselves properly in the DRE’s app folder
• Earlier, CUDA Toolkit was required to utilize CUDA, but DRE now cleverly downloads only a small portion of the CUDA package.
• Now, without the foundation for using CUDA, DRE won’t start downloading various files for CUDA when not necessary.

✅ Bugs caught by the internal bug tracker