PromptFlow Voice
The first voice app that auto-translates your words into a formatted, professional output — and injects it directly into whatever app you're typing in.
No credit card required · 7-day free trial · Windows 10 & 11
Start Free Trial →PromptFlow Voice is an AI voice dictation app for Windows. Hold your hotkey, speak naturally, and it detects the app you are using — then transforms your speech into the right format and injects it directly where you are typing. No copy-paste, no cleanup.
PromptFlow adapts your words based on where you are typing. The same spoken sentence produces a different output depending on the app that is focused.
Full enhancement, grammar-only, raw transcription, translated enhanced English, or translate-only. PromptFlow can format for the app, or preserve the meaning without rewriting.
Real speakers get a formatted, ready-to-use English output injected directly into their app. No copy-paste. No rewriting. No English required.
Most knowledge workers lose 2+ hours a day to typing. PromptFlow Voice dictates at up to 150 WPM versus the average 45 WPM typing speed — over 3x the throughput on text-heavy tasks.
At a $35/h rate, 2h of typing per day, switching 65% of it to voice saves roughly $183/month in time value. Voice Pro is $12/month. Net saving: ~$171/month. PromptFlow Voice costs $12/mo. You calculate the rest.
Every feature below exists because raw transcription wasn't enough. Speak once; PromptFlow Voice handles the cleanup, the formatting, and the typing.
PromptFlow Voice detects which application you're dictating into and shapes the output to match. The same spoken thought becomes a structured prompt in ChatGPT or Claude, a send-ready email in Gmail, a precise instruction in VS Code, or a short direct message in Slack. You never reformat by hand.
Hold your hotkey, speak, release. PromptFlow Voice only listens while the key is held down — there's no always-on microphone, no wake word, and nothing recorded by accident. When the key is up, the mic is off.
Dictate in French, Arabic, Japanese, German, or 30+ other languages and get clean, professional English injected straight into your app. Built by a non-native English speaker tired of transcription tools butchering his accent — the AI layer corrects what the microphone mishears.
Brand names, code identifiers, acronyms, client names, technical jargon — PromptFlow Voice protects them exactly as written instead of "correcting" them into nonsense. Add your own vocabulary to the personal dictionary once, and every future dictation respects it.
Five modes, one keystroke apart: full enhancement for structure, grammar-only to clean up your own voice, transcription-only for exactly what you said, and two translation modes — polished or literal. Most dictation tools are all-or-nothing. This one isn't.
The finished text appears directly where your cursor is — email client, browser, code editor, chat window. No floating companion window, no clipboard step, no switching apps to collect your text. If you can type there, you can dictate there. Windows 10 and 11.
A side-by-side look at PromptFlow Voice versus Windows Voice Typing (Win+H) — what actually happens after you stop speaking.
| Capability | Windows Voice Typing (Win+H) | PromptFlow Voice |
|---|---|---|
| Raw speech capture | Converts speech into plain text. | Converts speech into text, then decides what the output should become. |
| Removes filler words | Types every "um," false start, and repeated word literally. | Strips filler, false starts, and self-corrections so the text reads like you meant it. |
| Focused app context | Does not know which app you are typing in. | Detects the app context and adapts the output for Gmail, AI tools, code editors, browsers, or chat. |
| Email formatting | Leaves you with a paragraph to clean up. | Can create a subject line, greeting, structured body, and send-ready tone. |
| AI prompt formatting | Still requires manual prompt engineering. | Turns rough ideas into structured prompts with role, task, constraints, and expected output. |
| Translation | No translation — transcribes whatever language you speak, as-is. | Supports translate + enhancement for app-ready English, or translate-only when you want the same meaning without rewriting. |
| Control level | All-or-nothing literal transcription. | Choose full enhancement, grammar-only, transcription-only, enhanced translation, or direct translation. |
| Your vocab is protected | Often mishears names, tools, acronyms, and domain terms. | Protects brand names, code identifiers, acronyms, jargon, and personal dictionary terms. |
| Final placement | Types raw text at your cursor — you still fix and reformat by hand. | Injects the finished, formatted text directly where you are typing. |
Dictation puts a microphone in your workflow. Here is exactly how PromptFlow Voice handles it — in plain terms, no fine print.
Start free, no credit card. Stay monthly at $12 — or skip subscriptions forever and own it for $49, once. (A year of Wispr Flow runs about $180.)
$0 / 7 days
Try every Pro feature free for 7 days. Download from the Microsoft Store — no credit card required.
$12 /mo
For users who want everything ready to use, with no daily limits to think about.
$49 once
Pay once, keep it forever. Subscription dictation apps run about $180 a year — this is $49 one time, using your own Deepgram and Gemini keys.
PromptFlow Voice is a standalone desktop product. Choose how much AI assistance you want each time you speak.
No credit card required · 7-day free trial · Windows 10 & 11
Or skip the subscription entirely — own it for $49, once
A year of Wispr Flow is ~$180
Free local dictation forever · 7-day Pro trial · No card · Windows 10 & 11
150 vs 45
words per minute — speaking vs typing
< 4 s
from key-up to landed text
30+
languages in, professional English out
5
output modes, one keystroke apart
Claude needs a structured prompt. Cursor needs a developer-level spec. Gmail needs a polished email. Slack just needs you to sound human.
Same brain. Same thought. Four different formats — and you’re the one doing that conversion by hand, all day.
PromptFlow Voice does the converting. You just say the thing.
REC · Hold to dictate — double-tap to go hands-free
YOUR PACE
Hold the key and say as much as you need — or double-tap it to lock hands-free and keep going. Fillers, restarts, accents, another language: all fine.
T+0.0s
The moment you let go, the clock starts. PromptFlow reads the app you’re in — Gmail? Claude Code? Slack?
T+2.4s
Filler stripped, tone matched, structured for wherever you’re typing.
T+3.9s
Finished, formatted text appears at your cursor. No copy-paste, no cleanup.
The same 8 seconds of speech, dictated into four different apps. PromptFlow Voice reads the context and formats accordingly — you never touch the output.
Voice input · 8 seconds
“my app is crashing when two users try to edit the same thing at the same time. i dont know how to handle that. i need to fix it so they dont overwrite each others changes.”
Implement a robust concurrency control mechanism to prevent application crashes and data loss when multiple users attempt to edit the same resource simultaneously. - Analyze the current data update logic to identify the root cause of the crash during concurrent writes. - Implement Optimistic Concurrency Control (OCC) using a versioning field or timestamp to detect and prevent "lost updates." - Update the backend to handle write conflicts gracefully, returning a specific error or status code (e.g., 409 Conflict) when a race condition is detected. - Ensure the frontend handles these conflict responses by notifying the user and providing a path to resolve the discrepancy (such as refreshing the data) instead of crashing. - Provide production-quality code with clear error handling and minimal unrelated refactors.
Built by a non-native English speaker tired of dictation tools butchering his accent. Speak in your language — the AI layer corrects what the microphone mishears and formats the output for wherever you’re working.
FR · French
mon api plante quand trop de requêtes arrivent en même temps
→ English outImplement rate limiting to prevent API crashes under high concurrent load.
AR · Arabic
الرمز بينتهي صلاحيته بعد ساعة والمستخدم بيتسجل خروج لوحده
→ English outImplement silent JWT refresh — detect expiry 60s before, renew in background, no user interruption.
DE · German
die bilder sind zu groß in der datenbank nach dem upload
→ English outResize uploaded images to max 800px width before saving. Maintain aspect ratio.
JA · Japanese
ログイン後にユーザーを前のページにリダイレクトしたい
→ English outStore pre-login URL in session, redirect user back after successful authentication.
ES · Spanish
el formulario no valida el email correctamente, deja pasar emails sin el arroba
→ English outImplement email validation on frontend and backend. Reject inputs missing @ or containing spaces.
ZH · Chinese
用户登录后我想把他们重定向到上一个页面
→ English outSave referrer URL before login redirect. Restore and navigate after successful authentication.
Hear it happen — full recordings, two languages
These two demos use AI-generated narration — German and Hindi aren’t languages the creator speaks — used purely to show the translation pipeline end to end. Dictating in your own voice performs the same or better; there’s no synthetic audio to smooth over in real use.
Most dictation tools are all-or-nothing. PromptFlow Voice has five modes, one keystroke apart — a heartfelt message stays yours, a work email gets the full polish.
Rough speech becomes polished, structured, app-ready output. The AI picks the right format for the app you’re in.
Your exact words, kept. Only grammar, clarity, and flow get fixed — nothing is restructured.
Precisely what you said, untouched. Pure speech-to-text, no AI involvement at all.
Speak French, Arabic, Japanese — 30+ languages in, polished app-ready English out.
Direct translation into English. Same meaning, your phrasing, no rewriting.
Most knowledge workers lose 2+ hours a day to typing. Drag the sliders and see what that costs in real money — and what you get back with your voice.
Monthly receipt — your time
Recognizes 100+ apps and sites — plus the browser tab, via the free Chrome companion — and shapes the output to match. Five automatic tone styles, from technical to casual, each overridable per context.
Hold your hotkey to dictate; release and it stops. Or double-tap it to lock recording on — talk as long as you want with no key held down. One more tap stops it. The mic is never on unless you asked.
Finished text appears where your cursor is — native insertion, with a clipboard fallback that restores your original clipboard afterward. No companion window, no app switching.
Dictate in 30 languages and get professional English injected straight into your app. The AI layer also corrects what the microphone mishears — including accent drift and stutters.
Full enhancement to raw transcription, one keystroke apart. You decide how much AI touches your words.
A local Whisper model ships in the box — dictation works with no internet, no account, no license. Five downloadable models (Tiny 78MB to Large 3.1GB), every download SHA-256 verified.
A draggable, always-on-top mic that expands into four quick toggles: AI enhancement, auto-translate, language selector, and polish. Clear visual states from idle to done.
A personal dictionary protects brand names, identifiers, and jargon exactly as written. Voice snippets expand a spoken trigger phrase into full text — addresses, sign-offs, boilerplate.
Words appear as you speak, not after you stop — live streaming transcription on Voice Pro and BYOK (via your Deepgram key). The free local tier transcribes each take after you finish instead.
The mic is pre-warmed, so there’s no lag on first press. A live waveform shows what’s being heard, one click cancels a take — and on Voice Pro, if the daily AI quota runs out mid-session, it falls back to local transcription automatically.
Win+H transcribes literally — every um lands in your text, and it has no idea what app you’re in.
| Capability | Win+H voice typing | PromptFlow Voice |
|---|---|---|
| App context awareness | Doesn't know which app you're in | Detects and adapts for coding, email, chat, Slack |
| AI prompt formatting | Requires manual prompt engineering | Turns rough ideas into structured prompts |
| Filler word removal | Types every um, false start, repeated word | Strips filler, fixes self-corrections |
| Translation | No translation — transcribes as-is | 30+ languages → polished English output |
| Output modes | All-or-nothing literal transcription | 5 modes from full AI to pure transcription |
| Direct injection | Raw text at cursor — you still fix and reformat | Finished formatted text injected directly |
Comparing against Wispr Flow instead? Read the full breakdown
The microphone activates only when you start a dictation — key held or locked hands-free. No always-on listening, no wake word.
Hosted audio and transcripts are deleted immediately after processing — never stored. On the free local tier, audio never leaves your machine at all.
On BYOK, your speech goes straight to your own Deepgram and LLM provider accounts. Your data, your providers.
API keys and license are encrypted with Windows DPAPI. Device identity is a one-way hash — never raw hardware identifiers.
Free browser companion
Our Chrome extension puts a searchable prompt library inside ChatGPT, Claude, and Gemini — with multi-step prompt chaining and doc-to-.json conversion. Run it alongside Voice and it reads your browser context more precisely, so formatting gets even more accurate. Completely free, works with or without Voice.
Local dictation is free forever — unlimited, offline, no account. Add hosted AI for $12 a month, or skip subscriptions and own it for $49, once.
Free Local
Local model only
Unlimited offline dictation with the bundled Whisper model. Free forever — no license, no account.
Download free7-day Voice Pro trial included — 15 AI calls/day, no card.
Voice Pro
Managed hosted access
Keep everything running — we handle the infrastructure. No API keys, nothing to set up.
Buy Voice ProPaid subscription. Cancel anytime.
Everything in Free, plus:
Power User BYOK
Bring your own API keys
Same app, your own API keys, no monthly fees. A year of Wispr Flow is ~$180 — this is $49, once.
Buy license — $49Instant access. No subscription.
Everything in Voice Pro, plus:
Not sure yet? The free download includes a 7-day Voice Pro trial — 15 AI-enhanced calls a day, no card, no account. When it ends, unlimited local dictation stays free forever.
A year of Wispr Flow costs about $180. PromptFlow Voice BYOK is $49 — once.
REC · Ready when you are
Free local dictation forever. 7-day Pro trial — no card, no account. Windows 10 & 11.
A year of Wispr Flow is ~$180. PromptFlow Voice BYOK is $49 — once.
