Aqua Voice recently launched!
"Dictation as seamless as talking to a professional writer"
TLDR: Aqua is a voice-native text editor. Imagine dictating to a professional writer who can clean up your phrasing, fill in words, and understand your intent.
Founded by Finn Brown & Jack McIntire
🟩 Watch the demo video here ⤵️
🔴 Problem
Dictation still sucks:
- It's not very accurate.
- You have to learn specialized commands.
- Edits require the keyboard.
Finnian Brown has never been a very good speller (he’s dyslexic) and has been using dictation software since the sixth grade, when his dad got him set up on Dragon Dictation to write papers for school. It's always been easier for him to dictate—despite the janky software.
Dictation has remained surprisingly rigid in the era of LLMs. They make a lot of mistakes (see Siri lmao). All existing dictation services rely on learning a set of specialized commands.
They struggle to distinguish between filler phrases/commentary and stuff you actually want to write down.
🌎 Aqua
Aqua innovates on this paradigm by fusing real-time audio and an LLM to understand your intent. It does as much of the writing and formatting as you ask it to.
Accuracy is King 👑 - They use an MoE transcription architecture to have a lower WER than Whisper-largev3 in real-time.
Think Don’t Type 🧠 - Aqua can execute arbitrary commands. You don’t need the keyboard. Most people talk twice as fast as they can type.
Make Writing Chill 🏝️ - Aqua reacts naturally to feedback and can recover from mistakes if you tell it to!
