voxtty blog
Voice dictation, local-first tooling, and getting words out of your head and into your editor without touching the keyboard.
Whisper transcribes ~99 languages out of the box on Linux — the real constraints are model size, low-resource languages, and code-switching.
DictationMost people get comfortable with voice dictation in about a week — fluent, edit-free drafting takes longer, and no local model "learns" your voice.
GuidesMost dictation errors trace to mic gain, room acoustics, or model size — not the software. Here's how to fix each on Linux.
ComparisonsPer-minute cloud dictation pricing quietly adds up for writers and support staff who talk for hours — here's what it costs, and what "free" costs instead.
DictationWhisper transcribes natural-language commit messages and comments well, but stumbles on camelCase, variable names, and symbols — here's the practical split.
PrivacyDictating credentials is riskier from the room you're in than the software you use — here's when it's fine and when it isn't.
DictationVAD detects speech start and stop by analysing audio, not a countdown — here's how it works and why fixed-timeout dictation tools feel clunky.
ProductivityVoice dictation can reduce keystroke volume, offering a practical tool to manage RSI and typing fatigue alongside ergonomic practices, without cloud privacy concerns.
ComparisonsWindows Voice Access and macOS Dictation have deep OS integration Linux lacks — here's what that costs you, and what local-first tools trade for it.
ProductivitySpeech runs 130-150 wpm to typing's ~40, but editing overhead changes the math — here's where dictation wins and where it doesn't.
LinuxA practical walkthrough of the pieces needed for global voice dictation on Linux: ydotool, a hotkey daemon, PortAudio capture, and a systemd user service.
PrivacyThe real reason dictation software went cloud-first was GPU cost, not accuracy — here's what changed and what "process and discard" actually means for your audio.
ComparisonsOn-device Whisper models match cloud dictation for clear speech but lag on heavy accents, noise, and jargon — here's the honest breakdown.