JustScribe
On-device instant voice transcription
85 followers
On-device instant voice transcription
85 followers
JustScribe is a privacy-first live transcription app for macOS. Instant, offline speech-to-text powered by AI. No cloud, no data collection. Your voice, your data.
SquircleJS
I recently started vibe-coding by talking. Unfortunately, I'm not a fan of uploading my voice to the cloud AND paying for that every month. That's why I've build JustScribe! 🎉
I hope you find it useful, and as always - any feedback is welcome! Would mean the world to me if you share it on twitter!
@bring_shrubbery Streaming on-device transcription is harder than it looks. Whisper chunks in 30-second segments, so making JustScribe feel instant takes serious buffering work. Keeping voice data fully offline is the right call... voice is basically a non-resettable biometric.
SquircleJS
@piroune_balachandran Yeah, and also Parakeet models are insane and run really well on M-series processors, so it's just a win-win from users perspecive 👌
Product Hunt
Many people already use built-in macOS dictation or other tools like @Wispr Flow and @Aqua Voice; what’s the clearest scenario where JustScribe feels meaningfully better within the first 5 minutes, and what would have to be true for a user to switch and stick?
SquircleJS
@curiouskitty Nr. 1 reason is that it's offline and free. I'm building it mainly for myself, so the user experience will improve with time. macOS dictation is already worse at transcribing than JustScribe. @Wispr Flow does work better because of grammar correction, but this feature will also be added soon to JustScribe, and no subscription will be needed to use it 💙
Thanks for your question btw:)
Free and offline already wins the pricing fight against Wispr Flow at $144/year. Grammar correction is the gap though, and doing it locally means running a second model on top of Whisper. If JustScribe can nail that pipeline without tanking latency, the paid tools lose their strongest argument.
SquircleJS
@piroune_balachandran I will try my best, and I do believe that Parakeet + Llama model combo can really achieve this. I will launch again once that feature is ready, so feel free to follow 💙
Is there any way to make the trigger button just function? That's my biggest reason to use Spokenly - it's free (well, dirt cheap with API usage) and I can activate it with the function key that is so close to where my hand is resting, ha.
SquircleJS
@george_esther Yep! Me too, especially when vibe coding 😅 Too bad I can't talk in every situation 😭
It’s super cool Antoni! Wish you all the best here
SquircleJS
@german_merlo1 Thank you! 💙
I just love the fact that it's a no data collection tool. Love this, and congratulations on the launch, @bring_shrubbery !
SquircleJS
@neilverma Yeah, I hate the idea of uploading my voice to a random server :D Thanks, and I hope you enjoy using it!