I’ve decided to shift the focus of Vorn a bit—from purely a speech transcription app to something that lets you interact with your tools more naturally.
We’ll be launching Vorn Notes, which automatically saves notes to apps like Things and Obsidian.
– Jude
Building in public
Claude has really good models but something has to change. It is 1:00 not even during peak hours and we are having issues with Claude code and it won't even do simple things for me. Something needs to change, the models, infrastructure and or pricing.
- Jude
Building in Public
F*uck it VORN Voice, The fastest Speech to Text on the market is now 1$ for the first 25 people to download. This is a one time payment and it gets you a lifetime access to the app and all feature updates going forward. Use code VORN85 at the checkout of vorn.sh 😎
- Jude
Building in Public
Automatically Mutes your speaker so if you are listening to music and then want to speech to vorn it will automatically mute so you can speak.
- Jude
Building in Public
New Vorn Voice Desktop APP Icon
V1.0.6 for Mac OS, Windows Build coming soon.
- Jude
Building in Public
Big update to our app's payment structure! 🎉
Coming in the next few days!
💰 One-time $6.99: Full app + all local models (no cloud)
🔄 $6.99/month: Everything + cloud features
One time and monthly both include all future app updates
More flexibility, same great experience!
- Jude
Building in Public
Grinding out some new features today! Excited to see it all coming together.
- Jude
Building in Public
Vorn Voice V1.0 is out!
Press a hotkey, speak, get text pasted anywhere.
New in V1.0:
- Cloud transcription
- Local transcript polish
- Sidebar notes + Obsidian sync
- Themes
- Mac OS & Linux
Download for free for a limited time at vorn.sh/beta
- Jude
Building in Public
Loving the cloud model on Vorn Voice, its fast and just works!
- Jude
Building in Public
New Youtube Video! | Vorn Voice
youtu.be/AmAHARD5Ig8
- Jude
Building in Public
New themes coming to VORN VOICE as well as discord coming soon!
- Jude
Building in Public
What is stopping people from signing up for vorn? Is is the price or the amount of value coming from the product?
V0.1.12 VORN VOICE DROPPED
- Jude
Building in Public
New re themed website. Check out HTTPS://vorn.sh and sign up for the mailing list. I only send out emails about new features and updates and I do not use AI.
- Jude
Building in Public
Had a bug where custom key-binds wouldn’t update from default. Claude and GPT-5.4 couldn’t fix it after a long time I thought might as well try good old GPT5.3 codex and it solved it instantly. Insane.
- Jude
Building in Public
Big changes to Vorn Voice coming soon, Working on a new UI with other new features including a quick note taker that auto saves to obsidian. We are getting ready for an official 1.0 release pretty soon and I am excited for what is coming
- Jude Building in Public
Check out this menu bar for claude.
github.com/sk-ruban/not...
Exactly, it removes simple words like umms and ahhs and just cleans it up a tinny bit. Adding a dictionary feature next that allows people to add custom words. Any other ideas?
Exactly that's the goal, I am hopping to use a small model like the whisper tiny or base and work on fine tuning it for English and then I might have a cloud feature for vorn that allows for more voice options
Polishing has been officially released in V0.1.11. The first time you run a transcription it takes a little bit of time but after that its insanely fast. Working on a few different features which will make us stand out. Excited for what is coming!
- Jude
Building in Public
Text Polishing almost ready adding a few more things before I release a new version. If you don't already go grab a basic plan from vorn.sh and try out Vorn Voice. New secret features coming soon to vorn. What an amazing day to be alive!
- Jude
Building in Public
I am still working on it, they seemed to be all okay. But I think my benchmark is not standard enough so I am moving to a more popular standard one that I am going to try again. All of the whisper models are around 10% or less Word Error rate so if I can keep that and be faster ill be happy
- Jude
Is there anything specific people want in an ADE or AI tools in general, I want to broaden what I am working on into something a little bigger than just a transcription tool.
- Jude
Building in Public
Official VORN.sh website just dropped if you had an older version of the app you will NEED to re-download from the new site. I am going to be opening up a limited beta tester program soon
- Jude
Building in Public
I was able to double the inference speed and get the overall error rate down. I was fine tuning the tiny whisper using a small dataset of 30 clips around from 10 different languages. It would train for 15 minuets to see if its better Then auto research changes the training algo and tries again
This is what training with auto research does. Ran 40 tests overnight and these are the findings from the initial whisper tiny model to my new and improved model. Also remember these tests are running on a 10 year old server and are a lot faster on a mac book.
- Jude
Building in public
Exactly thanks for this!
Exactly thanks for this. I am hopping to keep the accuracy at or above 90% which is better then most of the whisper models while trying to get the inference down as much as possible
