Handy
An offline, cross-platform speech-to-text app that pastes local Whisper transcriptions into any active window.
Transcribe your voice locally with custom shortcuts and paste text directly into any application.
Handy is an offline speech-to-text desktop application that uses local Whisper and Parakeet models to transcribe your voice directly into any active text input. If you need local dictation for notes, scripts, or audio logging without cloud API fees or privacy issues, this tool fits your workflow.
What it is
The application runs in your system tray and listens for configurable global hotkeys to capture audio using Silero Voice Activity Detection. It processes your voice on-device using local GPU or CPU inference engines and pastes the transcribed text straight into your focused desktop application.
Who it's for
It suits audio producers, podcasters, and developers who want fast voice dictation without transmitting sensitive voice recordings to third-party cloud servers. It is less suitable for users who require real-time multi-speaker diarization or web-hosted team collaboration features out of the box.
Key features
Local Speech Recognition
Transcribes audio locally using Whisper GGML/GGUF models or CPU-optimized Parakeet V3 models.
Global Keyboard Shortcuts
Triggers push-to-talk or toggle recording modes instantly from any desktop application.
Voice Activity Detection
Filters silent audio segments automatically using Silero VAD before running speech inference.
Raycast Extension Integration
Allows controlling recording states, switching models, and searching transcript history via Raycast.
Up and running in 3 steps
Our honest verdict
What's great
- Transcribes speech completely offline without cloud API requests or telemetry.
- Pastes transcribed text directly into whichever desktop input field is currently active.
- Includes remote control CLI flags for integration with external automation scripts.
- Provides customizable inference options between GPU-accelerated Whisper and CPU-friendly Parakeet V3.
Where it falls short
- Requires installing third-party Linux utilities like wtype or xdotool for reliable text injection on Wayland or X11.
- Using macOS Bluetooth headsets causes audio output quality to drop due to system-level bidirectional mode.
- Whisper model inference can crash on specific Windows and Linux hardware configurations.
- Fn key shortcuts on macOS only register when using native Apple hardware keyboards.
Pricing & license
Handy is completely free to use and extend under the MIT license. There are no hosted cloud tiers or software subscription fees, so your only cost is the hardware resources needed to run local models.
Open remote AI-training roles
870 open remote AI-training roles, updated November 14, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Does Handy send voice audio to any external cloud servers?
No, all voice detection and speech-to-text processing occur completely on your local machine.
Which speech recognition models can I use with Handy?
You can choose between local Whisper models (Small, Medium, Turbo, Large) or CPU-optimized Parakeet V3.
Why does text paste fail on Linux Wayland desktop environments?
Wayland restricts automatic key pressing, so you must install wtype or dotool manually for paste functionality.
