Frequently Asked Questions
🧠
Smart Prompt Engines & Dynamic Personas
Rusper doesn't just transcribe raw speech; it refines your spoken stream-of-consciousness through specialized LLM prompt personas. You can switch personas with one click in the Dashboard or create your own custom prompt:
- •💻 Developer & Technical Specification: Preserves code signatures, camelCase, JSON formats, wraps variables in backticks (
`userId`), and detects spoken length/expansion cues. - •🔥 Smart Self-Correction & Emotion Stripping (Banger): Erases verbal venting, frustration, and throat-clearing while seamlessly resolving mid-sentence plan revisions.
- •✉️ Professional Email & Workplace Message: Formats spoken rambles into clean, structured corporate emails or Slack updates with proper paragraphs, bullet points, and greetings.
- •📝 Executive Summary & Action Items: Condenses meetings and brain dumps into scannable markdown bullet points.
- •✍️ Clean Verbatim: Minimal polish that fixes spelling, punctuation, and stutters while keeping 100% of your exact spoken words intact.
- •⚙️ Custom Persona: Write your own system instructions to format text however your specific workflow demands.
✂️
Voice Smart Editing
Voice Smart Editing is Rusper's superpower for rewriting existing text anywhere in Windows without manual copy-pasting:
- Highlight any text in any application (VS Code, Word, Chrome, Slack, Notion, etc.).
- Press your Smart Edit hotkey (default Ctrl + Alt + Space or custom key).
- Speak your instruction (e.g., "translate this to Spanish", "make this punchy", "turn into bullet points", "fix spelling and grammar").
- Release: Rusper sends your selection + voice instruction to Groq, transforms the text, and instantly replaces the highlighted selection in-place.
📚
Custom Vocabulary & App-Aware Context
⌨️
Global Hotkeys & Native Windows Injection
🎙️
Operating Modes & Glassmorphic HUD
⚡
Sub-300ms Speed, Privacy & $0 BYOK Model
Rusper runs on Groq's specialized Language Processing Units (LPUs). Groq LPUs execute OpenAI's Whisper Large v3 Turbo model in the cloud at ~216x real-time speed, giving you instantaneous transcription faster than local GPUs and without heating up your computer.