Deepgram is cloud (not local, not OSS) but their free tier is basically unlimited for personal use, so $0 forever.
Want fully local? Whisper is open source and runs 100 % on your Mac.
Tiny/faster models are just one line in the code away — anyone can add them in 30 seconds (PR welcome )
Both options free, zero data sent if you pick Whisper.
hey HN,
watched another “voice AI” startup raise $81 million for a fancy wrapper around whisper + gemini.
nah.
open-sourcing the one i built for myself instead → jarvis.ceo
30-sec demo: https://jarvis.ceo
repo: https://github.com/Akshayaggarwal99/jarvis-ai-assistant
hold Fn → talk → release → text + punctuation appears instantly
100% local or bring your own keys
zero tracking, zero subscription, zero bullshit
download the dmg or fork it and go wild.
i’m just one guy with insomnia and a macbook. if it saves you 5 minutes today, a star would make my entire month
thanks for looking
akshay
Jarvis: A native voice command layer for your desktop
I’ve been hacking away at Jarvis on nights and weekends for the past few months. I found myself increasingly frustrated with the constant context-switching involved in using ChatGPT through a browser for quick AI tasks. So, I decided to build a native voice-command layer that works seamlessly within any app you’re currently using.
It’s still pretty early, but I set up a frictionless web demo at jarvis.ceo/demo, where you can test it instantly without installation or sign-up. I’d love your feedback!
Want fully local? Whisper is open source and runs 100 % on your Mac. Tiny/faster models are just one line in the code away — anyone can add them in 30 seconds (PR welcome )
Both options free, zero data sent if you pick Whisper.