Beyond Flow
Voice dictation that runs on your own graphics card. Speak, and it types into whatever window you were working in, cleaned up or translated as you go. It can also hand your words back as a voice note, in an AI voice, or in your own cloned voice.
Requirements
- Windows 10 21H2 or later
- An NVIDIA graphics card with current drivers
- A microphone
- About 9 GB of disk space for the app and its speech models
Installing
See Installing a Beyond application for what the installer does and where it puts things. Your computer checks the download for you, and that page also shows how to check it yourself if you would rather.
Features
Dictation that stays on your machine
The large, accurate speech model runs on your own graphics card and the audio never leaves the machine. No account, no server, nothing metered.
One shortcut, any window
Hold to talk, or tap twice and talk hands free while each finished phrase appears as you keep speaking. The text goes wherever you were typing: an editor, a browser, a spreadsheet or a chat.
Cleaned up as you speak
Grammar and punctuation are fixed on the way, so what lands on the page reads like you wrote it.
Speak one language, type another
Choose the language you want out and speak normally. It translates as you go.
Text, a voice note, or your own voice
Your words can land as text, be saved as a voice note, or be read back in an AI voice or in your own cloned voice.
It learns your words
Correct a transcript once and it remembers the fix. Paste last brings back your latest dictation if it did not land.
Known limits
- Reading notes aloud in Google voices, if you choose them, sends that text to Google. Your dictation itself never leaves the machine.
Licensing
Beyond Flow uses the standard Beyond licence: checked on your own machine, with your devices managed by you. See Licences and activation.