Voooxly - Not voice-to-text: speak the task, get it done. Free, local

Hey Product Hunt 👋 Here's something I actually said out loud this morning: "um so I was thinking maybe we push the launch to friday, no wait, thursday works better, and uh let's tell the team in the standup" Here's what appeared where my cursor was: "I think we should push the launch to Thursday. Let's tell the team in the standup." No editing. No cleanup pass. That's Voooxly. It's not voice-to-text. Voice-to-text hands your words back to you. Voooxly hands you the task, finished. You pick a mode before you speak, and what lands at the cursor is done: the reply written and ready to send, the meeting summarized in bullets, the ramble turned into a structured LLM prompt, the paragraph in English, the spec written up from a spoken description. 9 modes, which is really 9 tasks. (Verbatim is in there too, for the rare day a raw transcript is what you want.) That's where the time actually goes. Talking beats typing, sure. But the real saving is the pass that never happens: no rereading, no punctuating, no reformatting before you hit send. Under the hood: whisper.cpp (large-v3-turbo) runs on-device, so your audio never leaves the machine in any configuration, because there's nowhere for it to go. The step that turns words into the finished task runs on a local Ollama, or your own Claude / OpenAI / Gemini / Grok / Groq key. And if you don't have any of those: get a free Groq key. It takes about a minute, no credit card, and it's what I've been using myself for weeks. It's fast and it works perfectly. Free is genuinely enough here, not a crippled tier I'm nudging you off. Whichever you pick, I never see the key. The part I didn't expect to love most: when it mishears a name, you just fix it in the text field like you would anyway. Voooxly notices and has the spelling right within seconds, before you've even left the field. No word lists, no dictionaries, nothing to maintain. You correct it once and that's it. The honest bit: free, MIT, no account, no telemetry, and no pricing page quietly appearing in month three. Apple Silicon (M1+) and macOS 13+ only for now, no Intel or Windows yet, and it's push-to-talk rather than live captions. Hold a key, say what you mean, let go. That's the whole product. AMA, I'll be here all day 🎙️ Edu