Universal-3 Pro by AssemblyAI — Speech-to-text that finally understands context
Speech-to-text that finally understands context
Promoted
🧐 Good find
Fn doesn't work for PTT, and I can't seem to change the hotkey either (I did get to the settings for it, but nothing happens box prompting for me to enter new keypress)
@veronism Hotkey logic has gottten a bit too complex. Might have found why that is happening. Shipping a fix in a couple hours - app will will auto-update.
This issue should now be fixed. Much more reliable hotkey binding now released in 1.7
If the app was open, it will auto-update. You can also force an update by right-clicking the menu button. Thanks for taking the time to report the bug!
@heathcliff1 You can just turn off the Wi-Fi and check :) I didn't believe it the first time the model worked too
Report
I love that you made it a one-time payment based. How do you plan to stay above breakeven when LLM credits usage will be too much? I think in 2-3 years timeframe you probably will lose money on LLM credits
@pasha_tseluyko thanks you, Pavel. Apple chip is powerful enough to transcribe voice to text. With more on-device formatting options in the works
Report
Love the offline-first approach. Curious - how does it handle longer recordings? I've found some dictation tools start lagging after a few minutes of continuous speech.
Fn doesn't work for PTT, and I can't seem to change the hotkey either (I did get to the settings for it, but nothing happens box prompting for me to enter new keypress)
Speakmac
hey @veronism , that's not expected. Can you relaunch the app once? (cmd+Q and launch again)
Speakmac
@veronism Hotkey logic has gottten a bit too complex. Might have found why that is happening. Shipping a fix in a couple hours - app will will auto-update.
Speakmac
This issue should now be fixed. Much more reliable hotkey binding now released in 1.7
If the app was open, it will auto-update. You can also force an update by right-clicking the menu button. Thanks for taking the time to report the bug!
I am not sure if it truly runs locally though
Speakmac
@heathcliff1 You can just turn off the Wi-Fi and check :)
I didn't believe it the first time the model worked too
I love that you made it a one-time payment based. How do you plan to stay above breakeven when LLM credits usage will be too much? I think in 2-3 years timeframe you probably will lose money on LLM credits
Speakmac
@pasha_tseluyko There is no LLM cost. Everything runs on-device efficiently with full accuracy
@kiranjd wow Kiran, you surprised me very much. How do you do it?
Speakmac
@pasha_tseluyko thanks you, Pavel. Apple chip is powerful enough to transcribe voice to text. With more on-device formatting options in the works
Love the offline-first approach. Curious - how does it handle longer recordings? I've found some dictation tools start lagging after a few minutes of continuous speech.