Push-to-Talk Dictation on Windows: The Complete Setup Guide
Quick answer: Push-to-talk dictation on Windows means holding a hotkey (Lightning Assist's default is Ctrl+Win), speaking while it's held, and releasing to have the transcribed text typed into whatever's focused — your mic is only ever active for the seconds you're holding the key. Setup takes about five minutes: install, enable AI Speech on the Triggers page, pick a microphone, and start dictating in any app. Windows also ships a built-in option, Voice Typing (Win+H), which is a genuinely useful toggle-based tool — but it works differently from true push-to-talk, and this guide covers exactly where each one fits.
If you've searched for this, you've probably already tried talking to your computer and had it either mishear you constantly or pick up half a conversation you didn't mean to dictate. Push-to-talk exists specifically to fix that second problem.
Push-to-Talk vs. Always-On Dictation
The distinction matters more than it sounds like it should. An always-listening microphone is constantly processing audio, which means it's also constantly at risk of picking up a phone call, a conversation with someone walking past your desk, or a video playing in another tab — and typing fragments of it into whatever you happen to have focused. It's also, for a lot of people, simply uncomfortable: a microphone that's always live is a microphone you have to remember is live.
Push-to-talk flips the model. The microphone is off by default and only activates while you're physically holding a key down. Release the key, and it stops — immediately, with nothing left running in the background. That gives you three concrete things always-on dictation doesn't:
- No false positives. Nothing gets transcribed unless you deliberately triggered it.
- Predictable privacy. You always know exactly when your mic was live, because you were the one holding the key.
- A workflow you can rely on under pressure. Press, speak, release, check the text. Same four steps every time, in every app.
Setting Up Push-to-Talk on Windows With Lightning Assist
- Download and install. Get the Windows build from /downloads — the installer takes under two minutes, and the 14-day trial requires no credit card.
- Sign in and open the Triggers page. Inside the app, navigate to Triggers and toggle on AI Speech. This is the push-to-talk feature; it's off by default until you enable it.
- Confirm your hotkey. The default push-to-talk shortcut on Windows is
Ctrl+Win. You can rebind it from the same page if it conflicts with something else you use — an IDE shortcut, a window-snapping shortcut, or another app's global hotkey. - Pick your microphone. Under Settings → Audio, choose your input device. A USB headset or a dedicated desk mic will noticeably outperform a laptop's built-in mic, especially in a room with any background noise.
- Test it. Click into any text field — a Notepad window is fine for a first test — hold
Ctrl+Win, say a full sentence, and release. The transcribed text should appear at your cursor within a second or two. - Use it for real. Once it's working in Notepad, it works identically in Outlook, Slack, your terminal, VS Code, or any other app that accepts keyboard input — because the tool operates at the OS level rather than inside any one application.
That's the entire setup. There's no per-app configuration step, because push-to-talk isn't integrated into any specific application — it types into whatever has keyboard focus, the same way your physical keyboard does.
Windows' Built-In Voice Typing (Win+H) — What It Actually Does
It's worth being precise about Windows' own tool here, because it's genuinely good at what it does and gets recommended constantly without much explanation of how it actually behaves.
Pressing Win+H on a hardware keyboard opens Windows' built-in Voice Typing, which uses online speech recognition powered by Azure Speech services — meaning it requires an active internet connection to work at all. Once activated, it listens for a "Listening…" indicator and then transcribes your speech into whatever text field currently has focus, anywhere on the system with a text box. Microsoft's own documentation notes that "fluid dictation will be automatically disabled on secure fields like passwords or PINs," and that voice-editing commands (things like "delete that" or "select that") are currently only available in US English, even though the dictation feature itself supports more languages.
The important distinction for this guide: Win+H is a toggle, not a hold-to-talk control. Press it once, and the microphone stays live continuously — you can keep talking across multiple sentences without holding anything down — until you pause long enough for it to stop listening, or you toggle it off yourself. That's a different interaction model from true push-to-talk, where the mic is only ever active for the exact duration you're holding a key. Neither model is wrong; they're built for different habits. If you want to dictate a long paragraph hands-free without holding a key the whole time, the toggle model is more comfortable. If you want the microphone to be provably off the instant you're not actively speaking to it, hold-to-talk is the tighter control.
When to Use Each
Use Windows' built-in Voice Typing (Win+H) when: you want a zero-install option that's already on your machine, you're dictating a longer passage and don't want to hold a key the whole time, or you're on a machine where you can't or don't want to install additional software.
Use a dedicated push-to-talk app when: you want the microphone provably off except for the exact seconds you're speaking, you want the same hotkey and workflow on macOS or Linux as well as Windows, or you want dictation paired with text expansion and AI rewriting in the same tool — Lightning Assist lets you dictate a rough draft with push-to-talk, then clean it up with an AI command or drop in a standardized snippet, without switching apps. Win+H is dictation only; it doesn't expand snippets or rewrite text.
Neither replaces the other completely. Plenty of Lightning Assist users still reach for Win+H for a one-off quick note, and reach for push-to-talk when they're dictating into a chat app, a commit message, or anything where they want tighter control over exactly what gets picked up.
Practical Setup Tips
Rebind if there's a conflict. Ctrl+Win is the default, but if it overlaps with a window-management shortcut or something your IDE already uses, rebind it on the Triggers page before you get used to the default — breaking a habit later is more annoying than picking a clean key now.
Invest in the microphone, not the software. Accuracy differences between tools are usually smaller than accuracy differences between a laptop mic and a $30 USB headset. If dictation still feels unreliable after setup, the mic is the first thing to check.
Speak in full sentences. Fragmented dictation — pausing mid-thought, restarting, correcting yourself out loud — produces fragmented, oddly-punctuated transcriptions. Modern speech models rely on sentence-level context to place commas and capitalization correctly; giving them a complete sentence gets a cleaner result than dictating word by word.
Say your punctuation when it matters. Sentence-ending periods are usually inserted automatically, but mid-sentence punctuation and line breaks — "comma," "new line," "colon" — generally need to be spoken explicitly if you need them to land exactly right.
Do a quick cleanup pass rather than aiming for a perfect first take. Dictating at natural speaking pace and then spending ten seconds fixing a name or a mis-heard word is still faster than typing the same passage from scratch.
Frequently Asked Questions
Does push-to-talk work in every Windows app, or just specific ones?
It works in any application that accepts keyboard input, because it operates at the operating-system level rather than as a plugin for one app — that includes Outlook, Slack, Teams, your terminal, VS Code, browsers, and word processors.
Is Windows' Voice Typing (Win+H) the same as push-to-talk?
Not exactly. Win+H is a toggle: press once to start listening, and the mic stays active continuously until you pause or toggle it off — it doesn't require you to hold a key down. True push-to-talk activates only for as long as you physically hold the hotkey. Both are useful; they just fit different habits.
Does push-to-talk dictation require an internet connection?
Yes, for Lightning Assist's AI Speech (it uses a cloud-based transcription model and consumes AI Credits) and yes for Windows' built-in Voice Typing (it's powered by Azure Speech services and explicitly requires a connection). Neither is a fully offline feature.
Can I use push-to-talk and text expansion together?
Yes — this is one of the practical advantages of a dedicated app over Windows' built-in tool. Dictate a rough draft with push-to-talk, then trigger a text-expansion snippet to add a standard prefix, signature, or template, or use an AI command to clean up tone — all inside the same app, in the same text field.
Related Reading
- Push-to-Talk Desktop App — the full feature overview across Windows, Mac, and Linux
- Speech-to-Text for Desktop — how the underlying transcription, latency, and privacy model works
- Push-to-Talk Voice Typing in Email and Chat Apps — cross-app dictation in Gmail, Outlook, and Slack specifically