You sound like yourself on a call. Your ideas move fast, your language is precise, and your English disappears into the work. Then you switch to Slack. Suddenly you're writing the same message you just said effortlessly, and it takes thirty minutes. Not because you lack vocabulary. Because you're re-reading it. Again. Because written English is where you feel like a smaller version of yourself.
The call-to-text gap
This isn't unique to non-native English speakers, but it hits differently when your accent doesn't match your thinking speed. On a video call with your US team, you're fast. You interrupt politely. You hold your ground in a disagreement. Your presence is solid enough that promotion conversations mention "exec presence" as a strength.
Then Slack. You draft a follow-up to your manager. You read it back. It sounds stiff. You rewrite. Better, but now it sounds like you're over-explaining. You rewrite again. This is the third draft. You're now reading your own words like a translator checking for errors, not like the person who thought them.
Maria, a senior account executive at a European SaaS company, calls this the "careful version" of herself. On a call, she closes deals. In Slack, she sounds like she's asking permission. The gap is real. It costs her.
Why Grammarly doesn't solve this
Grammarly catches grammar. Grammarly doesn't give you back your voice. It also doesn't give you back time.
If you're a non-native English speaker and you've used Grammarly, you know the pattern: it flags a sentence as "unclear" and suggests a revision that's more formal and more generic than what you wrote. Your language becomes safer. It also becomes less you. Grammarly optimizes for correctness, not for confidence or speed. The real tax you're paying isn't grammar mistakes. It's rewrite cycles.
Tools like Wispr Flow and Otter.ai assume you need help capturing speech. That solves a different problem for a different audience, people whose issue is getting words down fast, not people whose issue is that the words they get down sound like a translation of themselves.
The identity-protective tax
The behavioral pattern is consistent across Swedish PMs, German engineers, and senior professionals across Northern Europe: you speak English fluently and often better than native speakers (sharper vocabulary, more careful listening). Written English is where you feel the exposure.
This isn't imposter syndrome. This is a real cognitive cost. Speech has rhythm, speed, and interruption. It's harder to edit yourself live, so you commit faster. Writing has a pause before sending. That pause becomes a mirror. You read the sentence as if someone else wrote it, and you notice it sounds formal. Or stiff. Or like you're hedging more than you do on a call. So you rewrite.
Rewrite cost: six to ten minutes per message in Slack. One to three rewrites before sending. That's 30 minutes a day in large organizations where you write 10 to 15 async messages that matter (not "thanks!" or "lol, yes").
What changes when you draft by voice
Most of these rewrites happen because the first draft is correct but sounds wrong to you. This is the wedge. You don't need a grammar checker. You need to draft at the speed of thought and then refine once, not rewrite three times.
Drafting by voice removes the pause. You speak the message like you would on a call. The output is rough (speech is messier than writing), but it's in your voice. Then a single pass to tighten it. Send. Sixty seconds instead of thirty minutes.
Recitey runs Whisper locally on your device with zero variable cost, no word limit, and no metering. This matters because you're not going to dictate if every message is tracked or charged. You draft in Slack, email, browser, terminal, every Windows app via the clipboard. One language, no translation layer. The rough draft turns into a clean sentence in under two seconds.
For Maria, this is the difference between finishing her follow-up email before her next meeting or opening it tomorrow morning and rewriting it. Between async Slack messages that sound like her and messages that sound like she's being careful.
Who this solves for
If you're a native English speaker, you don't experience this. If your written English already sounds like you, this isn't your problem. If you've used Google Translate or a translation extension, you know it makes this worse by erasing your voice entirely.
This is for people who are excellent in English and feel small in writing. It's for the PM who leads sharp calls but sounds hesitant in writing. For the engineer who explains complex systems clearly on a call but takes twenty minutes to draft a comment in an issue. For anyone who keeps a hidden rule: "On calls I'm myself. In writing I'm a safer version."
The shift
You're not trying to sound more English. You're trying to sound like yourself without the translation delay. That delay is mostly friction, not accuracy. Removing it means you stop being the watered-down version of yourself in writing. You sound like the person your team hears on calls.
The performance review that said "exec presence on calls but could be more concise in writing" starts to look like what it actually is: a measurement of how tired you are by the time you finished the third rewrite, not a judgment on your actual communication. Write by voice and the gap closes.