Every feature runs locally on your machine. No accounts, no cloud processing, no data collection. Just fast, private dictation.
Whisper runs on your machine. Nothing leaves your device.
WisperCode uses OpenAI's Whisper models directly on your hardware. Choose from tiny to large depending on your accuracy needs and available resources. GPU acceleration supported but not required.
Hold, toggle, press, or double-press to record.
Hold your hotkey to record, release to stop. Or toggle on/off. Or single-press. Or double-press. Configure the shortcut and the mode to match how you work.
Text formatting adapts to the active application.
WisperCode detects which app is focused and applies appropriate formatting. Casual style for Slack, formal for email, code-friendly for your IDE.
Strips ums, uhs, and false starts automatically.
Natural speech includes hesitations. WisperCode removes filler words, repeated starts, and verbal tics so the output reads cleanly without manual editing.
Teach WisperCode your technical terms.
Add custom words, acronyms, and proper nouns to improve recognition accuracy. Framework names, medical terms, legal jargon — give Whisper context and it performs better.
Expand short phrases into full text blocks.
Define abbreviations that expand into longer text. Say a trigger phrase and WisperCode inserts the full snippet. Useful for boilerplate, signatures, and repeated responses.
Punctuation, capitalization, and formatting handled.
Whisper handles punctuation natively. WisperCode adds sentence capitalization, custom dictionary replacements, and configurable post-processing rules.
Save and organize your transcriptions.
Every transcription can be saved as a searchable voice note. Browse by date, search by content, and export when needed. All stored locally in SQLite.