Settings, AI Models & Themes

⚡ This Machine

Settings → This machine checks every component the app needs and can install what is missing.

NVIDIA GPU acceleration: Speech transcription and the local AI use your NVIDIA GPU (CUDA / cuDNN) when it is available, and fall back to the CPU.
Hardware Encoders: NVENC (H.264 or HEVC), Intel Quick Sync, AMD AMF or CPU (libx264), with bitrate and preset set per stream key.
Windows audio capture: Capture microphones, desktop audio and the sound of individual applications.
Stream priority: The app raises its own priority only while a stream is going out, so a game keeps its share the rest of the time.

⚡ NVIDIA CUDA 🎬 NVENC / Quick Sync / AMF 🔊 WASAPI audio

🤖 Local AI

Highlight scoring runs on a built-in llama.cpp engine by default. Ollama and cloud providers are optional.

Built-in AI (llama.cpp): The default engine. It offloads the model to your GPU and frees that memory before heavy render and transcribe jobs.
Bring your own: Use Ollama, LM Studio or any OpenAI-compatible server, or OpenAI, Google Gemini or Anthropic with your own API key.
Speech transcription: Runs locally with faster-whisper, with CrisperWhisper supported as an alternative.
Local by default: With the built-in engine, transcription and scoring happen on your PC. Nothing goes to a cloud AI unless you pick one.

🦙 Built-in llama.cpp 🗣 faster-whisper ☁ Optional cloud keys

🎨 Themes

Pick the look of the app in Settings; it switches on the running window, no restart needed.

20 Palettes: Catppuccin Mocha, Tokyo Night, Dracula, Nord, Gruvbox Dark, Solarized Dark, One Dark, Synthwave, Carbon, Ocean Blue, Bubblegum, Emerald, Tangerine, Grape, Catppuccin Latte, Sunshine, Clear Sky, Fresh Mint, Blush and Lavender.
Live switching: The theme is applied to every open tab of the running app and remembered for next time.

🎨 20 Built-in Themes

Settings: this machine, AI models and themes