AI Dictation App

An AI dictation app you actually want to use.

Shruvi is a free AI dictation app for macOS and Windows. Hold a hotkey, speak naturally, and Shruvi turns your voice into clean, punctuated text — then pastes it into whatever app you are using. Powered by Whisper-class speech-to-text and AI cleanup so you never have to add commas by hand again.

macOS 13+ · Windows 10/11 · Free · No credit card

Why Shruvi for AI dictation

  • Whisper-class transcription — Groq's whisper-large-v3-turbo by default, near-human accuracy on clear speech.
  • AI cleanup with gpt-4o-mini adds punctuation, drops filler words, and tightens phrasing without rewriting your meaning.
  • Personal dictionary so the model gets names, jargon, and acronyms right after the first correction.
  • Cloud-backed AI runs equally fast on any laptop — your local hardware doesn't bottleneck quality.

How it works

  1. Install Shruvi from shruvi.com/download/mac or shruvi.com/download/windows.
  2. Grant microphone access and pick your hotkey in Settings (default Command + Shift + Space on Mac, Control + Shift + Space on Windows).
  3. Hold the hotkey and speak naturally — release when you are done.
  4. Within about a second, the AI-cleaned text is pasted into the focused app.

What you can do with it

  • Write emails three times faster — let the AI add the punctuation and structure.
  • Reply in Slack, Teams, iMessage, and Discord without breaking flow.
  • Draft long-form docs, blog posts, and meeting notes in Notion, Google Docs, or Word.
  • Dictate code comments, commit messages, and AI prompts directly into VS Code, Cursor, or Claude.
  • Fill out CRM entries, support tickets, and web forms hands-free.
  • Speak journal entries, brainstorms, and to-dos directly into your notes app.

Privacy

Audio is captured locally, streamed over an encrypted connection to the Shruvi backend, transcribed, and discarded as soon as the text is returned. Raw audio is never retained on Shruvi's servers and is never used to train any model — neither Shruvi's nor its providers'. If you are signed in, transcript text may be stored on your account so history and sync work across devices, and you can delete it at any time. Full details on the Privacy Policy.

FAQ

What AI models does Shruvi use?

Speech-to-text is Groq's whisper-large-v3-turbo by default, with OpenAI Realtime as an alternative. AI cleanup of the transcript runs on gpt-4o-mini. The provider is set on the server, not in the desktop app.

What does the AI cleanup actually do?

It adds punctuation, removes filler words like "um" and "uh", fixes obvious self-corrections, and tightens phrasing without changing your meaning. It does not rewrite or summarize your speech, and you can disable it in Settings if you want raw transcripts.

Does Shruvi run AI on my device or in the cloud?

Shruvi runs transcription and AI cleanup in the cloud, so it works equally well on a low-spec laptop and a fast desktop. An internet connection is required. For fully local AI dictation, a Whisper-on-device app like SuperWhisper or MacWhisper is a better fit.

Is it accurate enough for real work?

Yes. whisper-large-v3-turbo delivers near-human accuracy on clear speech. Add custom words, names, and acronyms to your personal dictionary to improve recognition for terms the model would otherwise miss.