A minimal speech-to-text widget that works locally on your Mac. Activate with a shortcut, speak anywhere you'd normally type, and watch your words appear instantly. Perfect for AI chats, emails, and any text field.
This is the 2nd launch from Harker. View more
Harker 2.0
Launching today
Harker turns your voice into text without sending a byte of your voice to the cloud.
The core is free and runs 100% on your Mac.
Upgrade to Premium when you want AI to rewrite, summarize, or translate what you just said.
Free Options
Launch Team / Built With
Harker
Hey Product Hunt 👋
I'm Manuele, the maker of Harker. I shipped v1 last June because I was tired of speech-to-text apps that streamed my voice to someone else's server.
Harker transcribes 100% locally on your Mac. Nothing leaves your machine unless you explicitly ask for it. Your voice never does.
With v2, I'm doing something that scared me for months: I'm making the entire local transcription engine free. Forever. No trial, no nag, no feature gates on the private core.
Here's the logic. Local transcription is the thing people need to trust. Trust is built by giving it away. What I charge for is Premium ($5.75/month): cloud-based AI that takes your transcript and rewrites it, summarizes it, translates it, turns it into an email, a bug report, a Linear issue. That part genuinely costs money to run, and it's opt-in every single time. Your default stays private. Your voice, your biometric data, never leaves your machine.
A few things I'd love feedback on:
1. Which text transformations would you want in Premium next? I have a shortlist but I'd rather hear yours.
2. If you're a v1 user: the migration path. Tell me where it hurts.
Launch-day offer for the PH community: 40% off Premium annual with code PHLAUNCH40, valid 72 hour.
I'll be here answering every comment today. Ask me anything 🎙️
Private STT in 2026 needs a clear answer to one question: does Harker run the model fully on-device, or is it local recording with cloud transcription? Those are very different privacy stories. Where do you land on this, especially for anyone handling client calls or medical notes. and also, congrats and good luck on the launch :)
Harker
@keith_hiyamojo Thanks, and great question! Harker runs the transcription fully on-device. The model is downloaded to your Mac and inference happens locally. Your audio never leaves the machine. No cloud transcription, no network request, nothing. Works in airplane mode :)
The only time anything touches a server is if you explicitly use a Premium transformation (like rewrite or translate), and even then it's only the transcribed text that gets sent, not the audio. Your voice, the biometric part, stays on your device no matter what.
For client calls or medical notes, that's exactly one of the use case I had in mind. Fully local transcription layer, no third-party processor in the chain.
And thank you!
@mnlcpc "Works in airplane mode" is the whole proof, everything else is marketing. Medical and legal have been waiting years for this, go well!
ran into this building with voice input — the moment you route audio through a cloud stt service, you're adding a data processor to your privacy policy whether you want to or not.
local changes the equation: no dpa, no "what are they doing with this audio," no explaining to users why their voice ends up on someone else's server.
the only missing piece for teams is windows — the compliance problem is platform-agnostic.
Harker
@webappski really well put, that's exactly why I built it this way. Once audio touches a cloud service, you've got a whole new layer of compliance to deal with.
With Harker, your voice never leaves your Mac. That's the part that matters most . Voice is biometric data, and it stays on your device no matter what. The free version works fully offline, and of course, no account needed. For Premium features, only the transcribed text gets sent for AI transformation, and nothing is stored.
Windows is in the works, should be available relatively soon.
I've been using Harker for nearly one year now and have had a great experience so far! I love how easy it is to just talk when prompting AI, instead of writing everything down. Also very happy with how responsive Manuele has been, I've shared feedback multiple times and got back great responses!
Harker
@milhoornaert Hey Mil! Great to see you here, thank you!
You had me at “works in airplane mode”. I’d love to see something like this on iOS.
Harker
@anthony_sanna1 working on it ;)
Harker has been a great app to use especially with long form typing! I find it to be very accurate and a better deal than others out there.
Harker
@bigbreakmusic Thank you Scott
Epsilla (YC S23)
@mnlcpc Congratulations. And happy product launch.
Harker
@huisong_li Thank you Huisong