Voice typing app: what to install on each device

Phones already ship with a voice typing app

The boring truth is the best voice typing app for your phone is the keyboard it came with. On Android, Gboard does it: tap a text field, tap the microphone at the top of the keyboard, and speak when it says "Speak now". You can speak punctuation out loud, like "period", "comma", and "new paragraph", though Google notes punctuation and talk-to-text aren't available in every language. On the iPhone, the keyboard has a microphone key that does the same job.

That covers texts, emails, search boxes, and notes, which is most of what anyone writes on a phone. A typical phone dictation is an email reply of 50–150 words, about 30 seconds of speaking; the built-in microphone handles that without complaint. The third-party keyboard apps crowding the store results wrap that same microphone in different branding, sometimes with a subscription attached. Before installing one, try the button you already have. This is the rare product category where the free, preinstalled option is also the sensible one.

One honest disclosure, since this is the point where a vendor would pivot to their mobile app: we don't have one. Whisper by Remskill is a desktop app for Windows and macOS, full stop. If your dictation life happens on a phone, the built-in keyboard is my recommendation, and you can close this tab with my blessing.

Desktops are where the built-in tools run out

Both desktop operating systems include voice typing, and both are worth knowing before you install anything. On Windows 11, press Windows key + H and a small voice typing widget appears. It needs three things: an internet connection, a working microphone, and your cursor parked in a text box. It supports 40+ languages and can insert punctuation for you if you switch that setting on. The internet requirement is the line to remember. Win+H is great right up until the hotel Wi-Fi isn't.

On a Mac, Dictation starts from the microphone key in the function row, a keyboard shortcut, or Edit > Start Dictation. There's no length limit, but it stops on its own after 30 seconds of silence, which in practice means it quits while you stare at the ceiling choosing a word. It punctuates supported languages without being asked, and Apple states that Dictation isn't available in all languages or regions.

Neither tool is bad, and both are fine for a sentence or three. The boundaries show up when the writing gets real. Custom vocabulary for your product names and clients isn't on either feature list, and neither is AI cleanup of the "um, so" layer your speech carries. On a Mac you can check whether dictation runs on-device; on Windows, the tool goes dark the moment you're offline. Those boundaries are the reason desktop voice typing apps exist as a category.

What a desktop voice typing app should actually do

The mechanism is simple to state. A desktop voice typing app sits in the background, watches one global hotkey, records while you hold it, runs the audio through a speech-recognition model, and pastes the result wherever your cursor is. Each part of that sentence is a thing to check before you install.

The hotkey matters because it makes dictation a reflex instead of a feature you go visit. The "wherever your cursor is" part matters because writing happens in email, Slack, a CRM, a code editor, and a browser form; a tool that types only in its own window is a notepad with a microphone. The model matters twice: once for accuracy, once for where your audio goes. Local models run on your machine and work offline. Cloud models send audio to a server and often transcribe better. A good desktop app lets you pick per situation instead of deciding for you.

Speed belongs on the checklist too, in plain numbers. On an M1 MacBook Air running a small local English model, Whisper goes from key release to pasted text in about 1.4 seconds; in cloud mode on a decent connection, about 1.1 seconds. A response in that range keeps dictation in the "reflex" category. If a tool makes you wait, open a window, or copy text out of a panel, the friction eats the speed advantage you came for.

Here is my one opinion for this page: the best productivity hack is fewer steps, not faster steps. Most tools try to speed up typing. Voice typing deletes it. The flow goes from "stop, sit, type" to "speak, done", which is how speech at 145 words a minute beats typing at 40 without anyone learning a new skill. An app that adds steps back has missed the point of its own category.

For a deeper tour of the category itself — engines, accuracy, the local-versus-cloud trade — we keep a separate explainer on voice typing software. This page stays on the narrower question: what to install, per device.

How Whisper handles voice typing on Windows and Mac

Whisper by Remskill is our answer for the desktop half of the question. You hold one hotkey (Ctrl+Space on Windows, Command+Option on a Mac), speak, release, and the text lands wherever your cursor is, in any app. While you talk, a small overlay floats on screen so you can see it's listening.

Under the hood you pick between three paths, and the app doesn't pick for you.

• Local Whisper — English-optimized and multilingual model families, from a ~140 MB Base model up to the ~3 GB Large v3. The multilingual family covers 90+ languages including auto-detect. The English-only models are exactly that, English only, so pick the multilingual family if you switch languages mid-day.
• Local Parakeet — Runs 5–10× faster than Whisper on CPU and covers English plus 24 European languages. The fast lane for mostly-English dictation.
• Cloud (your own OpenAI key) — Connects to OpenAI with your own API key, using gpt-4o-mini-transcribe or gpt-4o-transcribe for the speech itself and gpt-5-mini for the optional AI cleanup.

Everything local is free: the models run on your device, work offline, send nothing anywhere, and no card is needed to sign up.

There's one more trick worth knowing: an AI keyword. Begin a recording with "Hey whisper" and the app triggers AI processing of the transcribed text instead of pasting it as-is. Dictation for the everyday sentences, an assistant for the moments you want the text reworked. Same hotkey either way.

The test I trust most wasn't a benchmark. I gave Whisper to my younger daughter, showed her once (press, talk, release, paste), and she dictated a 90-word email to her grandmother about a lost tooth and the tooth fairy's exchange rate, no follow-up questions. Two days later she reported that "the hotkey doesn't work in my drawing app", which is how I learned that normal users don't know what a hotkey conflict is. They just know it doesn't work. The customizable-hotkey settings shipped that night. If a seven-year-old can run the loop, the loop is simple enough.

The embedded app above is the real desktop frontend, not a screenshot; click around it. And to repeat the disclosure, because install intent deserves a straight answer: Whisper runs on Windows and on Apple Silicon Macs, and nowhere else.

A short checklist before you install anything

Installing software is a commitment, even when the software is free. Five questions sort this out faster than any review roundup:

• Which device do you write on, honestly? Mostly phone: keep the built-in keyboard dictation and spend nothing.
• Does it type in every app? Look for a global hotkey that pastes at the cursor, not a separate window you copy from.
• Can it work offline? Windows voice typing can't. Local models can. If your writing includes things that shouldn't visit a server, this question goes first.
• Which languages, counted per engine? 40+ for Windows voice typing, 90+ on Whisper's multilingual models, 25 on Parakeet, English only on the .en models.
• Did you fix the microphone first? A $20 USB mic does more for accuracy than any model upgrade. I build the software and the mic still wins. Nobody in this industry enjoys saying that out loud.

If you came here from the research angle — accuracy numbers, how the engines compare — the speech to text app explainer covers that side. The plans and what's in them live on the pricing page; the short version is that everything local costs nothing.

When to skip Whisper

If your dictation happens on a phone, skip us. We have no mobile app, and Gboard or the iPhone keyboard's dictation is free and fully adequate. If you write a sentence or two a day on a Mac, Apple Dictation is built in, starts from the microphone key, and punctuates for you.

If you need meeting transcription with multiple speakers and summaries, that's a different product category from writing-by-voice, and you should shop in that category instead. And if you're on Linux or an Intel Mac, Whisper doesn't run there, so the built-ins and browser tools are your honest options.

Twenty years ago, voice typing meant reading calibration scripts to your computer and hoping. Today the honest summary fits in two lines: your phone already does it, and your desktop can do it properly with one install. The microphone has been the fastest input device on your desk for a while now. Most of us just haven't pressed the key yet.