Every voice-to-text tool gives you a transcription. But raw transcriptions are messy — full of filler words, broken sentences, and casual phrasing that doesn't belong in a professional email or a polished document.

MetaWhisp solves this with four processing modes that let you control exactly how your speech becomes text. Whether you need verbatim accuracy for meeting notes or AI text cleanup for client-facing documents, there's a mode built for it.

The best part: Raw mode is completely free and runs 100% on your Mac. No internet, no API key, no limits. The AI-powered modes (Correct, Rewrite, Translate) use your own OpenAI API key at a cost of pennies per day.

Raw Mode — Verbatim Transcription

Raw mode gives you exactly what you said, word for word. No edits, no cleanup, no AI. It runs entirely on-device using WhisperKit and Apple's Neural Engine, so your audio never leaves your Mac.

This is the default mode and it is 100% free with no limits. No API key required, no internet connection needed. Speak, and the text appears.

Before (what you say):
"so um I was thinking we should probably like move the deadline to next Friday because um the design team needs more time"

After (Raw output)

so um I was thinking we should probably like move the deadline to next Friday because um the design team needs more time

Best for: Meeting notes, journaling, legal transcription, accessibility — anywhere you need an exact record of what was spoken. Also great for getting started with dictation on Mac before exploring AI modes.

Correct Mode — Remove Filler Words and Fix Grammar

Correct mode is where AI text cleanup begins. It takes your raw transcription and removes filler words (um, uh, like, you know), fixes grammar mistakes, and adds proper punctuation — all while preserving your original meaning and voice.

This is the most popular mode for daily dictation. You speak naturally, and the output reads like you typed it carefully. It requires an OpenAI API key and costs roughly $0.01 to $0.05 per day of normal use.

Before (what you say):
"so um I was thinking we should probably like move the deadline to next Friday because um the design team needs more time"

After (Correct output)

I was thinking we should move the deadline to next Friday because the design team needs more time.

Notice what changed: filler words removed, capitalization fixed, period added. But the sentence structure and your word choices stay the same. This is dictation with grammar correction — not a rewrite.

Best for: Slack messages, emails to teammates, quick notes, code comments, and any situation where you want clean text without changing your voice.

Rewrite Mode — Speech to Polished Professional Text

Rewrite mode goes beyond cleanup. It transforms your casual, spoken language into polished, professional prose. Think of it as a speech-to-text rewrite engine — you ramble naturally, and it outputs something you'd be proud to send to a client.

Same API key requirement as Correct mode, same cost. The difference is in the output quality: Rewrite restructures sentences, upgrades vocabulary, and produces text that reads like it was carefully drafted.

Before (what you say):
"so um I was thinking we should probably like move the deadline to next Friday because um the design team needs more time"

After (Rewrite output)

I'd like to propose extending the deadline to next Friday. The design team requires additional time to deliver quality work.

The meaning is identical, but the tone is completely different. Two clear sentences instead of one rambling one. Professional vocabulary. Confident phrasing. This is the power of AI-powered speech-to-text rewriting.

Best for: Client emails, documentation, blog drafts, proposals, and any professional writing where tone matters.

Translate Mode — Speak Any Language, Get English Text

Translate mode combines on-device speech recognition with cloud-based translation. Speak in any of 30+ languages — Spanish, Chinese, Japanese, Russian, French, German, and more — and get clean text output in English or your chosen target language.

The Whisper model handles speech recognition locally on your device, then the OpenAI API translates the transcribed text. Same API key, same low cost.

Before (what you say, in Spanish):
"Estaba pensando que deberiamos mover la fecha limite al proximo viernes porque el equipo de diseno necesita mas tiempo"

After (Translate output, English)

I was thinking we should move the deadline to next Friday because the design team needs more time.

Activate Translate mode with a double-tap of the hotkey. Speak in your native language, and the output appears in English — ready to paste into any app.

Best for: Bilingual professionals, international teams, language learners, and anyone who thinks faster in their native language but needs to write in English.

When to Use Each Mode

Pick the right voice-to-text processing mode for every situation.

Scenario Recommended Mode Why
Quick Slack message Correct Removes filler words, keeps your casual tone
Client email Rewrite Polished, professional output
Meeting notes Raw Verbatim record, no AI changes
Code comments Correct Clean but not overly formal
Blog draft Rewrite Transforms rambling into structured prose
Journaling Raw Captures your authentic voice
Multilingual team chat Translate Speak your language, output in English
Legal dictation Raw Exact words matter, no AI interpretation

Pricing: Transparent and Fair

MetaWhisp itself is completely free. There is no subscription, no freemium tier, no trial period. The app costs nothing to download and use.

Mode Cost Requires API Key Works Offline
Raw Free forever No Yes
Correct ~$0.01-0.05/day OpenAI key No
Rewrite ~$0.01-0.05/day OpenAI key No
Translate ~$0.01-0.05/day OpenAI key No

You bring your own OpenAI API key — MetaWhisp never touches your billing. Typical daily cost for AI modes is one to five cents, which works out to roughly $1-1.50 per month. Compare that to $8-17/month subscriptions from competitors, and the savings are obvious.

Read our privacy policy for full details on how we handle data (spoiler: we don't collect any).

How MetaWhisp Processing Modes Compare

Other voice-to-text apps for Mac offer similar features, but the approach differs significantly.

Feature MetaWhisp SuperWhisper Wispr Flow macOS Dictation
Processing modes 4 built-in Custom modes AI edit only None
Filler word removal Correct mode Yes Yes No
AI text rewriting Rewrite mode Via custom modes Built-in No
Translation 30+ languages Pro only No No
Free verbatim mode Raw mode, free Paid only Limited free Free
On-device option Raw mode Some models Cloud only Partial
Monthly cost $0-1.50 $8.49/mo Freemium Free

SuperWhisper offers custom modes where you can define specific AI prompts for different contexts — a flexible approach, but it costs $8.49/month and requires configuration. MetaWhisp's four built-in modes cover the most common workflows out of the box.

Wispr Flow has strong AI editing, but everything runs through their cloud servers. Your audio and text are processed remotely, which is a dealbreaker for sensitive work. MetaWhisp keeps Raw mode fully on-device and only sends text (never audio) for AI modes.

macOS Dictation offers no processing modes at all. What you say is what you get — including every filler word and grammar mistake.

Frequently Asked Questions

How do I remove filler words from speech-to-text transcription?

In MetaWhisp, switch to Correct mode. It automatically removes filler words like "um," "uh," "like," and "you know" while fixing grammar and adding proper punctuation. It uses your own OpenAI API key and costs roughly one to five cents per day of normal use.

Does MetaWhisp processing work offline?

Raw mode works 100% offline with no internet connection required. It runs entirely on-device using WhisperKit and Apple's Neural Engine. The Correct, Rewrite, and Translate modes require an internet connection because they use the OpenAI API for AI text processing.

How much does AI text cleanup cost in MetaWhisp?

Raw mode is completely free with no limits. For Correct, Rewrite, and Translate modes, you provide your own OpenAI API key. Typical cost is $0.01 to $0.05 per day for normal usage, which works out to roughly $1 to $1.50 per month — far less than subscription-based alternatives like SuperWhisper ($8.49/month) or Otter.ai ($16.99/month).

Can MetaWhisp translate speech from one language to another?

Yes. In Translate mode, you speak in any of 30+ supported languages and MetaWhisp outputs the text in English or your chosen target language. The on-device Whisper model handles speech recognition, and the OpenAI API handles translation. Activate it with a double-tap of your recording hotkey.

What is the difference between Correct and Rewrite modes?

Correct mode preserves your original meaning and phrasing but removes filler words, fixes grammar, and adds proper punctuation. Rewrite mode goes further — it restructures your speech into polished, professional prose. Use Correct for Slack messages, code comments, and casual notes. Use Rewrite for client emails, documentation, and professional writing.

Try All Four Processing Modes Free

Download MetaWhisp and experience AI text cleanup for speech-to-text on your Mac. Raw mode is free forever. AI modes cost pennies per day with your own API key.

Download for macOS

macOS 14+ · Apple Silicon · Free