Overview
Blurt is an advanced, high-performance desktop application engineered for seamless native performance, local-first data processing, and refined user experience.
Blurt is built on open-source principles to guarantee security, privacy, and full reproducibility. All sensitive data transformations run locally on your hardware with zero telemetry or unexpected cloud dependencies.
Key Technical Features & Capabilities
Works anywhere you can type — the transcript is pasted into the focused**: High-performance integrated capability designed for native Blurt workflow.
One key, no chords — dictation is triggered by a single lone modifier**: High-performance integrated capability designed for native Blurt workflow.
Polished in one step — each utterance rides to AssemblyAI's dictation API**: High-performance integrated capability designed for native Blurt workflow.
Styles — save up to four custom styles, each a short set of free-text**: High-performance integrated capability designed for native Blurt workflow.
Your vocabulary — key terms bias the spelling of names and jargon, and text**: High-performance integrated capability designed for native Blurt workflow.
Fast — transcription and cleanup share one round trip that typically**: High-performance integrated capability designed for native Blurt workflow.
Accurate —**: High-performance integrated capability designed for native Blurt workflow.
Multilingual — works in 18 languages, detected automatically, and you**: High-performance integrated capability designed for native Blurt workflow.
Architecture & Technology Stack
Primary Languages: Swift, TypeScript
Frameworks & UI Runtime: SwiftUI / AppKit, Electron / Node.js
Build Systems & Tooling: pnpm / npm
Supported Platforms: macOS 12.0+ (Apple Silicon M1-M4 & Intel x86_64), Windows 10/11 x64, Linux
System Requirements
| Specification | Minimum Requirement | Recommended Specification |
|---|
| Operating System | macOS 12.0 Monterey / Windows 10 x64 | macOS 14.0+ Sonoma / Windows 11 x64 |
| Processor Architecture | Apple Silicon (M1+) or Intel Core i5 x64 | Apple Silicon Unified Memory (M2/M3/M4) |
| Memory (RAM) | 4 GB Unified RAM | 8 GB+ High-Speed RAM |
| Storage Space | 300 MB available local disk space | 1 GB+ SSD storage |
| GPU Acceleration | Metal 2 / DirectX 12 compatible | Apple Neural Engine / Vulkan capable |
Installation & Setup Guide
macOS Installation (Recommended)
1Download the verified .dmg release installer directly from CookApps.
2Open the disk image and drag Blurt.app into your /Applications folder.
3Launch Blurt via Spotlight (Cmd + Space) or Launchpad.
Alternatively, install using Homebrew if available:
brew install --cask blurt
Windows Installation
1Download the official installer package (.exe / .msi).
2Run the executable and follow the setup wizard.
3Launch Blurt from the Start Menu or desktop shortcut.
Building from Source (Development Setup)
git clone https://github.com/AssemblyAI/blurt.git
cd blurt
pnpm install || cargo build --release
pnpm run build
How to Use & Workflow
1Application Launch & Workspace: Open Blurt to load the responsive native workspace.
2Configuration & Preferences: Press Cmd + , (macOS) or Ctrl + , (Windows) to configure hardware acceleration, custom directories, and display themes.
3Primary Operation: Drag-and-drop source files or input parameters directly into the central window.
4Execution & Export: Initiate the processing pipeline and verify output files saved directly to your chosen local directory.
Troubleshooting & Common Issues
Problem: macOS Gatekeeper displays *"App is damaged and can't be opened"* or unnotarized developer notice.
Root Cause: macOS quarantine flag set on binaries downloaded outside the Mac App Store.
Resolution: Open Terminal and run:
xattr -cr /Applications/Blurt.app
Alternatively, navigate to System Settings > Privacy & Security and select Open Anyway.
Problem: High memory consumption during batch processing operations.
Root Cause: Large media buffers retaining uncompressed state.
Resolution: Adjust batch concurrency in settings or process assets in focused batches.