EnviousWispr is a free, open-source AI dictation app for macOS that runs entirely on Apple Silicon. It transcribes voice locally using dual speech-to-text engines (NVIDIA Parakeet for English and WhisperKit for 99+ languages), optionally polishes text with on-device AI or cloud models, and auto-pastes clean text into any app. No cloud, no account, no subscription—audio never leaves your device. Built for developers, writers, founders, and healthcare professionals.
Full on-device transcription with Parakeet TDT (English) and WhisperKit (99+ languages) · Optional on-device AI polish with EG-1 model, Apple Intelligence (macOS 26+), or Ollama · Bring-your-own-API support for OpenAI and Google Gemini · Custom vocabulary, emoji insertion, transcript history · Sub-second latency (0.43s median for raw text, 1.5s with polish) · Push-to-talk, toggle, and hands-free dictation modes · Open source under GPLv3 license · No account, no subscription, no usage limits
AI-researched pricing — verify on the official site before subscribing.
Use it for
— Real-time dictation in emails, Slack, and docs
— Code comments and git commit messages
— First drafts and essay outlines
— Patient notes and medical documentation
— Hands-free productivity for long-form writing
— Multilingual speech-to-text with automatic language detection
Get the most out of it
01Set a global hotkey (default: Right Option) and hold while speaking for quick one-liners, or double-press for hands-free long-form mode
02Import contacts directly from your Mac to enable automatic name correction without sharing data
03Start with on-device EG-1 or Apple Intelligence for fastest, most private AI polish; only use OpenAI/Gemini if you need advanced cloud-based formatting
04Use Custom Words to teach the app technical terms, acronyms, or jargon specific to your domain (e.g., 'Kubernetes', 'React', medical abbreviations)
05Disable PostHog analytics and Sentry crash reporting in Settings if you want zero telemetry
Build the autonomous home assistant infrastructure before consumer robotics achieve mass adoption. Combines local LLM control with voice interfaces and cross-device orchestration to control robots, sensors, and traditional smart home devices.
How the workflow runs
01richard-companion — Deploy self-hosted voice-controlled home automation powered by local LLM
02enviouswispr — Enable sub-second on-device voice dictation for issuing commands to AI systems
03ailyn — Orchestrate AI intelligence across multiple devices and edge computing nodes in the home
04walkers2 — Integrate humanoid service robot for physical tasks once consumer models become available
05agent-native-cloud — Scale home AI infrastructure as more robotic and IoT devices come online
Physical AI and robotics are predicted to achieve breakthrough consumer adoption through home automation. This stack positions early adopters to control the entire ecosystem: voice input via EnviousWispr, local reasoning via Richard, cross-device orchestration via Ailyn, and future robot integration via WalkerS2. When consumer robots hit shelves, you'll have the control plane ready.
