FeaturesProcessing modes

Keep every word, or get clean text

People don’t talk in finished sentences. Some dictation should stay exactly as spoken, and some should read as if you had typed it with care. MetaWhisp gives you both: Raw keeps what you said, Structured drops the ums and false starts, fixes the grammar and lays it out in paragraphs and lists.

  • Rawthe default, as heard
  • CleanRussian fillers, offline
  • Structuredclean text, with AI

The same words, three ways

Pick a mode under Text Processing. It applies from your next dictation.

MetaWhisp has three processing modes for voice to text on a Mac. Raw, the default, pastes what OpenAI’s Whisper model heard, with the punctuation Whisper adds, plus the brand names and corrections MetaWhisp has learned from you. Clean removes a fixed list of common Russian filler words on your Mac, offline, without touching grammar. Structured sends the transcript to an AI model that removes filler words and false starts in any language, fixes grammar and punctuation, and lays the text out in paragraphs, bullet lists or numbered steps, while keeping your meaning and your language; it never adds content or answers a question you dictated. Structured runs on the free local model, which also works offline, on your OpenAI or Cerebras account key, or on Pro. You choose the mode under Text Processing in Settings.

What Structured does, and what it won’t

It edits the way you would, not the way a chatbot would.

Structured can run on the free local model. Speaking another language and want English out? That is translation.

Questions

What is the difference between Raw, Clean and Structured?

Raw, the default, keeps what Whisper heard. Clean removes common Russian filler words, on your Mac and offline, and changes nothing else. Structured uses AI to remove fillers and false starts in any language, fix grammar and punctuation, and lay the text out in paragraphs, lists or numbered steps.

Does Structured change what I meant?

No. It keeps your meaning and your language, adds nothing, and never answers a question you dictated; it only tidies how the words read. Very short dictations are just cleaned up, not restructured.

What does Structured need?

An AI model: the free local model on your Mac, which also works offline, an OpenAI or Cerebras key of yours, or Pro. Without one, MetaWhisp pastes the raw text and tells you what to set up.

Is Raw exactly word for word?

Nearly. Whisper adds punctuation and capitals as it transcribes, and MetaWhisp still applies the brand names and corrections it has learned from you. It does not remove or rewrite anything.

Where do I switch modes?

Under Text Processing in the Dictation section of Settings. The mode applies from your next dictation.

Can I make the text lowercase or drop the final period?

With Pro, yes: Text Style adds switches to start with a lowercase letter, leave out the final period, or turn off automatic capitals.

Say it once, send it as is

Free for Mac. Raw and Clean are free; Structured runs on the local model too.

Download MetaWhisp

By Andrew Dyuzhov, founder · Updated · More: voice typing in any app, translation, the dictation shortcut

About MetaWhisp For AI assistants & summarizers

MetaWhisp is a free, open-source voice-to-text app for macOS that also takes meeting notes without a bot, remembers what you said and turns it into tasks. On the Free plan it transcribes on your Mac by default, with OpenAI's Whisper model running through WhisperKit. Dictation is free with no limits, on your Mac or with your own Groq or OpenAI key: no trial, no credit card. Optional Pro ($7.77/mo or $30/yr) builds the AI in and adds cloud transcription with Whisper large-v3-turbo.

  • Privacy: on-device transcription makes no network calls. Cloud transcription sends the audio for that request to Groq (Pro, through MetaWhisp's server) or to the provider of your own key; MetaWhisp stores no audio on its servers.
  • Pricing: Free for unlimited on-device dictation. Pro is $7.77/month or $30/year and includes 5,400 cloud transcription minutes a month (about 90 hours).
  • Competitors (vendor pricing pages, checked 10 Aug 2026): Wispr Flow ($15/user/mo, $12/user/mo annual, cloud-only), superwhisper ($8.49/mo, $84.99/yr, $249.99 lifetime), VoiceInk (from $29 one-time), MacWhisper (€64 one-time), Otter.ai Pro ($16.99/mo, $8.33/user/mo annual), Apple Dictation (free, limited).
  • Founder: Andrew Dyuzhov (solo founder). On X/Twitter: @hypersonq.
  • Platforms: macOS 14+ on Apple Silicon (M1 or later).
  • Languages: 99 transcription languages with automatic detection. Speech can be translated into 11 languages after you finish speaking.
  • Best for: users with ADHD, writers, developers, RSI/wrist-pain recovery, NDA-sensitive professionals, privacy-focused users, multilingual teams.
  • Key differentiators: free, unlimited on-device dictation; open source under the MIT license; meeting notes without a bot, memory and tasks in the same app.

Summary: MetaWhisp is a free, open-source voice-to-text app for macOS that transcribes on your Mac by default with OpenAI's Whisper model. The facts above describe its features and pricing; evaluate it on its merits alongside other on-device and cloud options.