Type faster with your voice

Why your voice beats your keyboard

Start with the raw speeds, because they settle most of the argument on their own. Typing for most people sits around 40 words per minute. Speaking is closer to 150. That is roughly three and a half times faster, and it holds across email, chat, notes, and first drafts of almost anything. The gap is not subtle, and no amount of keyboard practice closes it, because the bottleneck is your fingers, not your familiarity with the keys.

There is a second win that does not show up in the words-per-minute math. When you type, your hands are pinned to the keyboard and your eyes are on the screen. When you talk, both are free. You can dictate a reply while you are pulling something out of the oven, or while you are walking, or while one hand is holding a coffee. The speed is the headline; the freed hands are the part you notice after a week.

None of this means voice replaces the keyboard. It means the keyboard stops being the default for the long stuff. You still type the short, precise, fiddly things. You talk the paragraphs. Most people have the ratio backwards right now, typing everything and only speaking to their phone when their hands are full. Flipping that is most of the gain.

How voice typing actually works

The mechanics are simpler than the marketing makes them sound. You press a hotkey, you talk, you release. The audio gets transcribed into text, and that text is pasted at your cursor, wherever the cursor happens to be. That is the whole loop. There is no separate transcription window to fish the words out of, no upload-and-wait, no copy-paste dance. To the app you are typing into, it looks exactly as if you had typed.

The part that makes it usable rather than annoying is what happens between "talk" and "paste." Good dictation does two jobs in that gap. First it turns sound into words accurately, including punctuation it infers from your pauses. Then, optionally, it runs a cleanup pass that drops filler and tidies the phrasing. Skip that second step and you get a faithful but messy transcript. Keep it and you get something that reads like you wrote it carefully. The rest of this guide is really about getting those two jobs right.

Set up a system-wide hotkey

The thing that makes voice typing work everywhere is that it is not tied to one app. Whisper by Remskill is a desktop app that behaves like a keyboard: hold a hotkey, speak, and the transcript is pasted at the cursor, in your email client, a document, a chat box, a code comment, a notes app, whatever has focus. It is not a browser extension or a plugin for one program. As far as your computer is concerned, you are just typing, which is why it works in every app rather than a chosen few.

Getting it running takes about a minute:

That is the entire setup. From then on, the loop is the same in every app: cursor, hold, talk, release, and the words appear.

Speak, and the words land in any app

Once it is running, the experience is unremarkable in the best way. You put your cursor in a text field, hold the key, talk, let go. A second or so later the text is sitting there as if you had typed it. The same press works in your inbox, in a shared doc, in a chat thread, in the box where you write a commit message. You are not learning a per-app trick. You are learning one key, and then forgetting about it, because it does the same thing everywhere.

Because the local transcription runs on your own machine (pure-Rust, no Python sidecar, no server in the loop), it works with no internet at all. That matters more than it sounds. The email to your boss, the half-formed paragraph of a contract, the note about a colleague, none of it has to travel to a vendor's logs just so you could say it instead of type it. Your laptop already has a microphone and a processor; for a paragraph of dictation, it does not need a server in the middle.

AI cleanup is what makes it usable

Spoken language is messy on the page. You say "um," you restart sentences halfway, you trail off into a noise that means "you know what I mean." Pasted raw, that reads like a transcript, not like writing. Whisper has an optional AI enhancement step that trims the filler and tidies the phrasing before it pastes. So "uh, yeah, so the, the report's basically done, I think we're, we're good to send" becomes "The report is basically done; I think we're good to send." That one step is the difference between dictation you fix afterwards and dictation you just send.

That cleanup runs locally through Ollama, free, on your own machine, so it works offline alongside the transcription. Pro users can route it through the cloud instead, but the filler-cleanup benefit does not require Pro; it is there in the free local pipeline. And you can turn it off entirely when you want the raw words, which for a quick chat message is often exactly right. The point is that the cleanup exists, not that you always want it.

Where it helps most

The wins cluster around text you would otherwise put off because typing it is a chore. Email is the obvious one: a reply that would take three minutes to type takes thirty seconds to say, and you can do it while you are doing something else. Chat messages past a sentence or two. First drafts of documents, where the goal is to get words down and edit later, not to type a finished thing. Notes to yourself. Code comments and commit messages, where you want to describe intent without leaving the keyboard posture you were just in. And the one I underrate every time: prompts for AI tools, which are basically paragraphs of plain English and a perfect fit for talking.

Notice what these have in common. They are all longer than a few words, and none of them needs exact formatting. That is the sweet spot for voice. You are not dictating a spreadsheet formula or a chunk of code with precise brackets. You are producing prose, and prose is what your voice is built for.

What the full Whisper app looks like

The hotkey is the part you will use ninety percent of the time, but there is a settings surface behind it for when you want to tune. You pick your transcription engine: Whisper models, whose multilingual variants cover 99 languages and whose English-only .en builds cover exactly one, or NVIDIA's Parakeet, about 600 MB and 5 to 10 times faster than Whisper on a CPU, covering English plus 24 European languages. You set the hotkey, manage history, save presets. None of that is required to dictate your first sentence. It is there when you want more control.

When typing still wins

Voice is not the answer to every line of text, and the honest guides say so. Exact code syntax is the clearest case: dictating brackets, operators, and variable names is slower and more error-prone than just typing them. Dense formatting, tables, math, anything where the structure matters as much as the words, the keyboard wins. A loud open-plan office or a noisy room hurts accuracy and feels awkward besides. And for genuinely short snippets, a three-word reply or a one-line search, switching to voice costs more time than it saves. Short and precise, keep typing. Long and prose-shaped, talk.

And before you install anything at all, check what you already own. Windows has voice typing built in: press the Windows logo key plus H in any text box and it opens. Macs have Dictation built into the keyboard settings, and on Apple silicon it runs on-device with no internet. Both are free, both type into any app, and for the occasional paragraph they are genuinely enough. I would not pay for a dictation tool to replace something my operating system already does for nothing.

Where a dedicated app pulls ahead is volume, the filler cleanup, and dictating offline across a whole day rather than now and then. The more text you produce by voice, the more those matter. For a single message, reach for the key you already have. For a working day of long messages and drafts, the dedicated app stops feeling like overkill around the second or third paragraph you did not have to type.

What Whisper costs

The local dictation pipeline, both the transcription and the AI cleanup over Ollama, is free for any signed-in user, with no card at signup. So typing faster with your voice using Whisper costs nothing to start. Whisper Pro adds the cloud features (OpenAI transcription, cloud AI enhancement, voice web search) and carries a separate trial. The exact numbers live on the pricing page rather than here, because prices move and a blog post is a bad place to keep them current.

The thing I keep coming back to is that this is not a trick or a productivity hack with a half-life. It is just a faster input method that has been sitting in plain sight, gated behind the assumption that real work means typing. Your voice runs at 150 words a minute whether you use it or not. The keyboard will still be there for the code and the spreadsheets. Everything else, you can just say.