Horse Whisprer

Horse Whisprer is offline dictation for macOS: press a hotkey, speak, press again, and your words appear in the app you were already using. It is built for people who write across many places every day: notes, docs, email, Slack, browsers, editors, support tools, issue trackers, and any standard macOS text field. Instead of opening a separate transcription app, uploading audio, waiting for a cloud service, then copying the result back, Horse Whisprer stays in the menu bar and turns speech into text right where your cursor is. The app records microphone audio locally, transcribes it on your Mac with a whisper.cpp-compatible runtime, and inserts the result through paste or direct typing fallback. Your dictation workflow can be as simple as Control + Option + Space, or you can configure another shortcut. When insertion is not the right behavior, use copy-only mode and keep the transcription on the clipboard. Horse Whisprer is especially useful when you want fast drafts without sending raw voice recordings to a cloud dictation service. Brain-dump a paragraph, reply to a customer, capture meeting notes, draft a README section, write a journal entry, or dictate a long message without leaving the text field in front of you. Key capabilities: - Local speech-to-text on macOS using Whisper models - Global hotkey and configurable dictation triggers - Automatic insertion into the active app via paste, typed fallback, or copy-only mode - In-app model picker and model downloads, from small fast models to higher-accuracy options like Whisper large-v3-turbo - Language auto-detect plus common fixed-language options - Start and stop feedback sounds, with optional audio ducking or muting while recording - Menu bar status, setup checks, permissions guidance, and diagnostics for text insertion - MIT-licensed app code, with official signed and notarized builds available separately Most dictation tools make you choose between convenience and control. Horse Whisprer is for Mac users who want both: a lightweight native workflow, local transcription, configurable models, and text that lands directly where work is happening.