superwhisper on Windows

What superwhisper is and who it's for

superwhisper is an AI dictation app. You speak, it transcribes, and it drops formatted text into whatever app you're in — email, a chat box, a code editor. By its own description it works system-wide across any application, handles punctuation and formatting on its own, and supports 100-plus languages with automatic language detection. It also transcribes audio and video files, not just live speech. That's a genuinely good feature set, and I'm not going to pretend otherwise.

The part superwhisper leans on hardest is privacy. Its Windows page says it straight: "On-device AI models. Your audio never leaves your machine and you don't need internet." That's the right instinct for a dictation tool, and it's the same instinct behind Whisper by Remskill. Anyone who's watched a cloud bill climb knows why local matters. (I once watched a team rack up a five-figure cloud-AI bill in one quarter, mostly from re-transcribing standup recordings four times because a "smart retry" was too eager. The CFO opened the dashboard mid-review and the room went quiet.) On-device dictation sidesteps that whole category of problem.

So who's it for. Someone who wants a polished, modern dictation app, likes the idea of picking from several AI models for cleanup, and either lives on a Mac or has come over to the new Windows build. superwhisper grew up on macOS, where it's a strong, well-regarded pick. The Windows app is the newer member of the family. That's not a knock — it's just where it is in its life, and it matters for the comparison below.

"Why isn't it on Windows" — and why that finally changed

For years the honest answer to "superwhisper for Windows" was "it isn't, it's a Mac and iPhone app." Mac-first is a common path for indie dev tools — one platform to support, tight integration with the OS, fewer things to break. Windows is harder. The input plumbing alone is rougher; I learned more about the Windows input-method framework than I ever wanted to, back when our own hotkey fired the stop-recording callback six times per keypress on machines with a language input enabled. It took a 300ms debounce to fix. Mac never had the bug.

That gap is what sent people looking for alternatives, and it's why this keyword still gets typed. Old habits, old articles, and the fact that the Windows build is recent enough that plenty of people haven't heard it landed. The mechanic both tools share is the simple part: press a hotkey, speak, release, and the words paste at your cursor. A small capsule shows up while you talk so you know it's listening.

With Whisper by Remskill the push-to-talk hotkey is Ctrl+Space on Windows, held while you speak, and it's changeable in Settings if it clashes with something. (My younger daughter once told me a hotkey "didn't work" in her drawing app. It was a conflict, not a bug — which is how I learned the average person has no idea what a hotkey conflict is. So now every hotkey is customisable.) If you've set up dictation on Windows before, this is the same muscle memory.

Setting up Whisper on Windows in two minutes

If you want to try the alternative alongside superwhisper, here's the whole setup. You need a Windows 10-or-newer PC, a working microphone, and a couple of minutes. The entire local pipeline is free for any signed-in account, with no payment method asked for at sign-up.

The slow part is the model download, not the setup. Everything else is the four steps above. Once it's running, dictating into a field stops being a typing task and becomes a talking task. The same is true after you install superwhisper — both tools are a hotkey and a microphone away from working everywhere.

How superwhisper and Whisper compare, honestly

Both run on Windows now, both transcribe on-device, both work in any app, both cover a long list of languages. That's a lot of overlap, and I'd rather concede it than pretend a chasm exists. Where they differ is in the edges. superwhisper offers a roster of cleanup models — its site lists choices like GPT, Claude, Llama, Grok, Gemini, and Mistral — and a system of custom "modes" for shaping tone. If you love having that many knobs, that's a real draw, and superwhisper genuinely does it well.

Whisper by Remskill is built a little differently. It doesn't pick a transcription engine for you — it puts three paths in front of you and lets you choose: a fast local engine, a multilingual local engine that can also translate to English, or a cloud option using your own OpenAI key. On Windows the push-to-talk hotkey is Ctrl+Space, customisable. The local pipeline — every local model, AI cleanup through a local model, history, presets, custom hotkey, three devices — is free for any signed-in account with no card at sign-up. The paid tier only adds the cloud surface. That's a different shape from a single free-tier-plus-Pro split.

The one honest caveat about superwhisper on Windows is maturity, not capability. The Windows app is newer than its macOS sibling, and newer desktop software on Windows tends to have rougher edges for a while — that's just the cost of the input-plumbing problems I mentioned earlier. If you want the version with the most road behind it, that's macOS for superwhisper and Windows-and-Mac-from-the-start for Whisper. If you're a happy Mac-superwhisper user curious about Windows, try it; just don't expect day-one parity with the platform it grew up on.

Local or cloud: which Whisper mode on Windows

If you go the Whisper route, try local mode first. Most of the things people dictate — an email, a half-formed note, a message you'd never want sitting in a vendor's logs — have no business leaving your laptop for one paragraph of text. If your PC is from the last few years, local handles everyday dictation without complaint, and cloud becomes the escape hatch rather than the default. Here's how the three paths differ, because the app makes you pick and I'd rather you pick well.

The choice comes down to speed, language coverage, or top-tier accuracy.

• Local Parakeet — NVIDIA's TDT engine, around 600 MB, and the fastest local option — 5 to 10 times faster than Whisper on CPU. Covers English plus 24 other European languages, 25 in total. No translate-to-English. If you mostly speak English or another European language, this is the quick, fully offline pick.
• Local Whisper — slower than Parakeet on the same machine, but the multilingual builds cover 99 languages and can translate to English. The English-only builds are English-only, not 99. Pick this for Chinese, Japanese, Korean, or any translation work, which Parakeet can't do. Default English model is around 480 MB.
• Cloud (OpenAI, BYOK) — best accuracy and web access, using your own OpenAI key billed straight by OpenAI. Transcription runs on gpt-4o-mini-transcribe by default. Needs internet, so it's the one path that leaves your machine. The cloud surface is part of Whisper Pro.

The boring truth is that for most of what people dictate on a Windows machine, local is plenty. Both local engines run fully on your own hardware with nothing sent to a server — the same on-device promise superwhisper makes. Cloud earns its place when you want top-tier accuracy on a hard recording or you need the model to pull a fact off the web mid-sentence. Start local, reach for cloud only when local leaves you wanting.

The AI cleanup pass that turns rambling into prose

Raw dictation comes out as a run-on. You say "okay so review the architecture doc tag it project alpha and remind me thursday," and that's the unpunctuated wall any speech engine hands you first. What you do with that wall is where the tools earn their keep.

superwhisper does cleanup through its model picker and custom modes. Whisper by Remskill runs an AI pass too: say the activation phrase "Hey whisper" and the text gets enhanced before it lands — filler words stripped, run-ons fixed, the spoken paragraph turned into something you'd actually send. On a local model that pass runs through Ollama, fully offline; in cloud mode it's gpt-5-mini by default. Either way the goal is the same — you talk loosely, the text arrives tidy.

The cleanup pass is the difference between dictation you have to fix and dictation you can send. It's also where the on-device-versus-cloud line matters most: a cleanup model that runs on your machine never sees a server, which is the whole reason to dictate locally in the first place. If you only ever need raw transcripts, you won't miss it. The moment you're firing off real emails by voice, it's the feature you stop noticing because it just works.

That same speak-then-clean flow is what makes voice faster than the keyboard in the first place — you can type faster with your voice in any app, so a long paragraph becomes a few spoken sentences instead of five minutes of typing.

If you're on a Mac

Be honest with yourself about your hardware, because the right answer changes. If you're reading this on a Mac, superwhisper is a strong pick and I'd say so to your face. It started on macOS, it's been polished there for years, and on Apple Silicon its on-device models run well. For a Mac user who wants a mature, native dictation app with a deep model picker, superwhisper has genuinely earned the recommendation. I'm not going to send you away from a good tool to score a point.

Worth knowing: Whisper by Remskill runs on Mac too, on Apple Silicon, with the same local-first pipeline and a command+option push-to-talk hotkey held while you speak. So the choice on a Mac isn't superwhisper-or-nothing — it's two solid local options, and which one fits depends on whether you want superwhisper's model-and-modes approach or Whisper's three-paths-you-choose approach. Both keep your audio on your machine. Try the one whose shape you like; neither is a mistake.

And if you're sending a 30-word text, don't install anything. macOS Dictation is built in, set up in System Settings under Keyboard, and on Apple Silicon it can process general text on-device. On Windows, the built-in Voice Typing bar opens with Windows key + H wherever your cursor is — it punctuates on its own and is fine for short bursts, though it routes through Microsoft's servers and needs internet, so it isn't an offline option. (If Win+H ever stops opening, the usual culprits are in this fix for Win+H not working.) Reach for a dedicated tool when the built-ins start hurting — long notes, offline privacy, or wanting the same hotkey everywhere.

If you came here from the old guard of dictation, the same comparison logic applies to the modern Dragon alternative — local, no training step, and built for the way people actually dictate now.

"superwhisper for Windows" used to be a dead end. It isn't anymore — the app is on Windows, it's good at what it does, and if it's what you wanted you should just go get it. The only reason to read past the install button is to make sure you picked the tool that fits your machine and your privacy bar rather than the one that ranked first. I dictated most of this comparison on a Windows laptop, with a tool that pastes wherever my cursor happens to be, then sent it off. The keyboard sat there unused, which is the entire point.