You explained it perfectly on the call. Sharp, present, no hedging. Then you drafted the same thing in Slack and it sounded smaller, more careful, less like you.
The 30-minute rewrite cycle
Maria is a senior account executive at a European B2B SaaS company. Spanish background. Eight years selling to US enterprise customers. On a video call with a prospect, she's sharp, clear thinking, quick pivots, confident. But watch her draft an async message to her sales team about the same topic. She types. She reads it back. It doesn't sound like her. It sounds smaller. She deletes half. Rewrites the other half. Reads it again. Sends. She's just spent fifteen minutes saying something she articulated in ninety seconds on video.
Her performance reviews capture this asymmetry perfectly. "Executive presence on calls," they note. "Could be more concise in writing." The feedback stings because she knows it's not about her English. She speaks English all day. It's the medium.
This rewrite tax isn't unique to Maria. On a call, you think and speak simultaneously. Your accent's just part of how you communicate. But async? The medium creates space for doubt. You're no longer under time pressure to respond. You've got room to re-read, to second-guess, to wonder if this sounds professional enough. Maria re-reads her own Slack messages three or four times before sending. Each read-through is a small crisis: is this too casual? Too formal? Does this sentence make me sound unsure?
If you're spending thirty minutes a day caught in this rewrite cycle, that's 130 hours a year. That's three full workweeks.
But the real cost isn't hours. It's the identity erosion. Over time, you learn to write smaller. You sand down the edges. You remove the directness, the humor that comes naturally in real time. You become the careful version of yourself.
Why existing tools don't solve this
If the problem were grammar, Grammarly would solve it. Grammarly's excellent at catching grammar mistakes. But it doesn't give you back your voice. It assumes you need basic language help, and you don't. You need your voice to survive the translation to async.
Translation tools make this worse, not better. They erase the accent that's perfectly fine on a call. They strip out the personality. They assume you need to sound "neutral" or "professional" or "fluent," when what you actually need is to sound like yourself.
Otter.ai captures what you say, but it requires cleanup and review. You're trading speech for more editing, not less. MacWhisper works on Mac, but there's no Windows equivalent. Dragon NaturallySpeaking exists, but it's oriented to people who can't type at all, accessibility software, not for knowledge workers who're perfectly capable of typing but tired of the rewrite tax.
The landscape assumes the problem is that you can't write English. The real problem is that you can write English, but you doubt yourself in async, and the tools that exist either make that doubt worse or ignore it entirely.
The moment voice changes everything
You're about to send a Slack message. Instead of typing, you speak it. You say it the way you'd say it on a call: naturally, at full speed, with the rhythm and emphasis that comes without thinking.
"We should table the feature discussion until we see how the customer tests the current version. If they hit blockers, we'll have live feedback to inform the priority."
You just spoke that in maybe twelve seconds. Now you'd usually rewrite it, swap "table" for "defer", change "blockers" to "issues", soften "priority" to something less direct. You'd read it back. Rewrite again. Send after the third or fourth read.
Instead, the speech model processes it. Under two seconds. The output's clean. Not perfect, but it's your voice, not an edited approximation of your voice. You read it once. It sounds like you. You send it.
Recitey runs Whisper, an open-source speech model, locally on your device with no internet required and no data leaving your machine. Whisper-large-v3, which Recitey uses, achieves 96.3% word accuracy on LibriSpeech, that's the benchmark dataset for speech recognition. You're running the most accurate version available. It processes speech in real time and returns clean text in 1 to 2 seconds.
Here's what makes this different: zero variable cost per word, no metering, no word limits, no subscription caps that punish you for speaking instead of typing. The tool works in Slack, email, browsers, your terminal, every Windows app via the system clipboard. You don't have to move to a special app to capture your voice. You capture it where you work.
That architectural choice, running locally, no caps, working system-wide, means voice writing becomes a default behavior, not a workaround.
Your voice survives.
What changes after adoption
The first thing you notice is the time. You're not spending the first thirty minutes of your morning caught in the rewrite cycle. Instead, you're clearing your Slack backlog, responding to async messages, getting on with your day.
But the deeper shift's less visible. Over a few weeks, your team starts hearing you differently in async. Your messages sound like you. Not like a carefully edited approximation of yourself. Not like someone hedging or worrying about how English sounds. Like the person they hear on calls.
That matters more than the time savings. It matters because identity compounds. Over months, when your async communication sounds like your real voice, two things happen: you doubt yourself less, and your team understands you better. Fewer clarification questions. Fewer "did they mean this?" moments. Just clearer, faster, more present communication.
For Maria, this shift's the invisible benefit. She's not reclaiming 130 hours a year (though she is). She's reclaiming her presence in her own writing.
Who this is for
This isn't for people who need grammar help. Grammarly's good for that. This isn't for people learning English; there're better tools for language learning.
This is for people who're already fluent, already sharp on calls, but become hesitant when they move to async writing. For people who know their English's good but feel exposed in Slack because you suddenly have time to second-guess yourself. For non-native speakers who're excellent in spoken English and tired of the hidden tax.
It's not the right fit for highly formal contexts. Legal documents, regulatory language, formal proposals, those still need human review and careful editing. But Slack was never formal. Internal communication was always conversational, meant to capture what you'd say if you had the person in front of you.
The sharp version of you doesn't need to disappear when you move from calls to Slack. It just needs the right medium to survive.