Everything you need.
Nothing you don't.
EdgeWhisper is designed to be invisible until you need it. One hotkey, speak, release. Your words appear — formatted, polished, ready.
Real-Time Transcription
Speak naturally and see your words appear in real-time. Sub-500ms latency powered by Voxtral Mini 4B from Mistral AI, running locally via MLX on Apple Silicon.
Works Everywhere
Inject text into any macOS app — VS Code, Slack, Terminal, Safari, Mail, Notes, Google Docs. If you can type there, EdgeWhisper works there.
100% On-Device
Your audio never leaves your Mac. The 4-billion parameter Voxtral model runs entirely on-device. Zero cloud calls, zero telemetry, zero accounts.
AI Auto-Edits
Automatic punctuation, smart capitalization, and filler word removal. Say 'um' and 'uh' — EdgeWhisper removes them automatically.
Developer-Ready
Smart terminal detection for 20+ dev tools. Works perfectly in VS Code, Xcode, iTerm2, Warp, and JetBrains IDEs with intelligent injection.
Apple Silicon Native
Built in Swift for macOS. Runs Voxtral Mini via the MLX framework on the Neural Engine. No Electron, no overhead — pure native performance.
Snippet Library
40+ built-in voice shortcuts. Say 'calendar' to expand your scheduling link. Create custom snippets with variable support.
13 Languages
English, French, German, Spanish, Italian, Portuguese, Dutch, Hindi, Arabic, and more. Automatic language detection powered by Voxtral's multilingual encoder.
Works where you work.
EdgeWhisper injects text directly into any macOS application. Native apps, web browsers, Electron apps, terminals — everywhere.
For Developers
Dictate code comments, PR descriptions, documentation, and Slack messages. Smart terminal detection means it works perfectly in your IDE and terminal. No more context-switching to type.
For Writers
Write blog posts, scripts, emails, and social content at the speed of thought. Auto-editing handles punctuation and formatting while you focus on ideas. Writer's block meets its match.
For Professionals
Legal professionals, healthcare workers, and consultants who need accurate, private dictation. On-device processing means HIPAA-ready compliance without extra infrastructure.
Your voice stays
on your Mac.
Unlike cloud-based alternatives that route your audio through remote servers, EdgeWhisper runs the full Voxtral Mini 4B model from Mistral AI directly on Apple Silicon's Neural Engine. Your words are processed locally — they never leave your machine.
No accounts. No telemetry. No cloud. The architecture makes it impossible to collect your data. Voxtral is open-source (Apache 2.0) — you can verify every claim.
Works Offline
No internet required after the initial model download. Process speech without any network connection.
No Telemetry
Zero analytics, zero tracking, zero data collection. We don't know who you are or what you say.
Local Storage Only
All data stays in your Mac's Application Support folder. Full export and deletion at any time.
No Accounts
No sign-up, no email, no login. Download, grant permissions, and start dictating.
HIPAA-Ready
Since nothing leaves your device, EdgeWhisper is inherently compliant with data privacy regulations.
Open-Source AI Model
Powered by Voxtral Mini 4B from Mistral AI, released under Apache 2.0. Transparent, auditable, trustworthy.
Built for speed.
Obsessively optimized.
Native Swift. Voxtral Mini 4B from Mistral AI. Apple MLX framework. Neural Engine acceleration. No Electron, no web views — pure macOS performance.
<500ms
Transcription latency
Configurable from 240ms to 2.4s — Voxtral's sweet spot
4B
Parameter model
Voxtral Mini from Mistral AI, running 100% on your Mac
<5%
Battery impact
Optimized for Apple Silicon Neural Engine via MLX
Framework
SwiftUI + AppKit hybrid. Native macOS UI with deep system integration via Accessibility API.
ML Engine
Voxtral Mini 4B Realtime from Mistral AI via Apple MLX. Causal audio encoder + sliding window attention for infinite streaming. Optimized for M1/M2/M3/M4 Neural Engine.
Text Injection
AXUIElement primary, NSPasteboard + CGEvent fallback. Works in 20+ terminal and IDE applications.
Storage
SwiftData for structured content, UserDefaults for preferences, file-based model cache. All local.
Your voice deserves
better than the cloud.
Download EdgeWhisper free. Powered by Voxtral AI from Mistral, running 100% on your Mac. 5 minutes a day, zero setup, total privacy. Join developers and professionals switching from cloud-based dictation.
macOS 14+ · Apple Silicon (M1/M2/M3/M4) · Voxtral Mini 4B · Apache 2.0 AI · ~3 GB model