Typing a two-line reply is fine. Typing a detailed email, a thorough Slack message, or a long Telegram thread is a different exercise entirely, and somewhere around the third paragraph, the keyboard starts to feel like the wrong tool for the job.
Typing a message from scratch averages about 19 words per minute versus roughly 150 words per minute for speaking it aloud — a gap of nearly 8x. That difference isn't just speed: it's attention. Typing splits focus between your idea and the mechanics of hitting keys, punctuating, and correcting typos, so long emails and messages suffer the most.
| Task | Speaking (dictation) | Typing (composing new text) |
|---|---|---|
| Average rate | ~150 words/min (conversational English) | ~19 words/min (composing new text, not copying) |
| Time to draft a 400-word email | ~3 minutes | ~21 minutes |
| Attention split between content and mechanics | None — offloaded to Hovor's formatting pass | Continuous — punctuation, capitalization, and typos compete with your thought |
Sources: speaking rate — Mental Floss, citing the National Center for Voice and Speech; typing-composition rate — Wikipedia, "Words per minute".
For short messages this overhead is barely noticeable. Scale up to a long email (a client update, a project summary, a detailed question thread) and the cost compounds. You lose the thread of what you were saying. You shorten ideas to avoid more typing. The message you send is a compressed, less precise version of the one you had in mind.
On a phone it gets worse. The keyboard occupies half the screen, autocorrect fights you, and holding the device while typing at length strains your wrist. Many people abandon a message halfway and send something terse instead, not because they had nothing to say but because saying it by tapping felt too laborious.
Raw speech-to-text output is unpunctuated and hard to read, but Hovor doesn't stop at transcription. A formatting pass runs immediately after your words are transcribed, adding punctuation, correct capitalization, and paragraph breaks so what lands in your email or chat field is already readable — not a transcript you have to fix before sending.
The cleanup pass handles the practical issues that make raw dictation unusable:
You do not need to speak in a formal, deliberate way. Speak as you normally would, and the cleanup step translates that into clean written text.
On macOS, Hovor's continuous mode dictates hands-free: trigger it once, and every pause in your speech becomes a chunk that's transcribed, formatted, and inserted where your cursor is, while the mic stays live for the next phrase. Because this mode runs on-device via Parakeet rather than through the metered cloud path, a single session runs up to 31 minutes regardless of subscription tier.
Trigger it via the hotkey or the menu bar button. Your content accumulates in the message or document as a series of clean insertions; you never touch the keyboard during the drafting phase.
This changes the writing experience for anything longer than a short reply:
On iOS, dictation works through the Hovor keyboard: you tap to record, speak, and the transcribed text is inserted via the keyboard extension.
Hovor applies one of three tone presets — Casual, Professional, or Concise — to the same spoken input during the formatting pass, so a Slack message and a client email can come from identical dictation but read in the right register for each destination. You pick the tone before speaking; Hovor doesn't rewrite your ideas, only the surface register.
Select a tone before you dictate, and the formatting pass applies it:
This is especially useful when switching rapidly between channels. You might dictate a quick internal Slack note in casual tone, then dictate a client-facing summary in professional tone, without recomposing either in your head; just speak naturally and set the tone for the destination.
Hovor pastes transcribed, formatted text directly into whatever app has focus — no copy-paste, no window switching. On macOS it uses system-level text injection into virtually any text field (Mail, Slack desktop, Notion, a browser compose box). On iOS it works through the Hovor keyboard extension, the same input surface as the stock keyboard, in any app that accepts text.
On desktop (macOS), Hovor pastes into the active application automatically using system-level text injection. This works with virtually any app that accepts text input: Apple Mail, Spark, Superhuman, Gmail in a browser, Slack desktop, Telegram desktop, Notion, Linear, any document editor.
On iOS, dictation works through the Hovor keyboard extension. Switch to the Hovor keyboard in any text field (the same fields you would use a normal keyboard in), tap to record, speak, and the text is inserted through the keyboard API. This works across email clients, messaging apps, browsers, and anywhere else iOS shows a keyboard.
You can also use custom snippets: saved expansions that you trigger by voice or shorthand. For messages you send repeatedly (standard sign-offs, common project updates, boilerplate acknowledgements), a snippet lets you speak one short trigger phrase and have the full block of text inserted, already formatted.
Free tier includes 2000 words per week via Hovor's cloud transcription. Pro removes the limit: $11.99/month or $89.99/year, up to five devices.
Dictation saves the most time for four groups: people who write more than a couple of substantial emails a day, mobile-first workers typing on a phone keyboard, non-native writers composing in a second language, and anyone with RSI or typing discomfort. In each case, the composition phase — normally the longest and most repetitive part of writing — moves from keyboard to voice.
People who write a lot of email. If your day involves more than a handful of non-trivial email exchanges, the time saving compounds quickly. Composing on voice and editing with hands is faster than composing entirely on keyboard for anything over two paragraphs.
Mobile-first workers. Drafting long messages on a phone keyboard is one of the more tedious parts of mobile work. The Hovor keyboard on iOS lets you dictate directly in any app (the message field in Gmail, the reply box in Telegram, the compose view in any app) without switching to a separate tool.
Non-native language writers. Writing at length in a second language is cognitively intensive. Dictating in your natural spoken cadence, then letting the cleanup pass handle the written grammar, reduces the load considerably. You spend your mental energy on the content, not on translating your thinking into correct written form.
Anyone with RSI or typing discomfort. Extended typing sessions are a known aggravator. Dictation removes the mechanical load for the composition phase, the part that is usually the longest. Editing and reviewing remain keyboard-based, but those tend to be shorter and less repetitive than composing from scratch.
Up to 31 minutes per session, regardless of whether you're on the free tier or Pro. Continuous mode runs entirely on-device via the Parakeet speech model, so it isn't subject to the 2-minute cap that applies to a single cloud-based recording on the free tier (Pro raises that cloud cap to 31 minutes too). If you're still talking after 31 minutes, stop and restart the mode — your in-progress text is already in your document.
Continuous, hands-free dictation is a macOS-only feature. On iOS, you dictate through the Hovor keyboard extension: switch to it in any text field, tap to record, speak, and the transcribed, formatted text is inserted through the keyboard API — one tap-record-insert cycle per phrase, rather than the always-listening loop continuous mode uses on Mac.
Hovor's free tier includes 2,000 words per week of cloud transcription, at no cost. Pro removes that limit for $11.99/month or $89.99/year and covers up to five devices. Two one-time purchases exist alongside subscriptions: Local Unlock ($49.99) for unlimited on-device dictation, and BYOK Unlock ($24.99) to use your own OpenAI or Anthropic API key; a bundle of both is $69.99.
It depends which speech and formatting path you use. On-device transcription (Parakeet, gated to iPhones with 4 GB+ RAM) works fully offline, and continuous mode on Mac uses this on-device path. The default cloud path (OpenAI transcription and formatting) requires an internet connection. Hovor does not currently offer a working offline local-LLM formatting model — the cleanup pass runs via the cloud, Apple's on-device Foundation models (iOS 26+), or your own BYOK provider.
Yes. Before dictating, choose Casual, Professional, or Concise. The formatting pass applies the selected tone during cleanup, adjusting register and trimming filler without rewriting your underlying ideas. You can switch tones between messages — dictate an internal Slack note in Casual, then a client email in Professional, without recomposing either by hand.
Download Hovor for iOS or macOS. Free tier includes 2000 words/week; Pro unlocks unlimited dictation across up to five devices.
Get Hovor