Clatter Labs
Console

Line CLI

Voice and privacy

Speak your prompt in Arabic or English with Ctrl+Space. What is sent, what is kept, limits and microphone setup.

Speak your prompt instead of typing it. Line turns your speech into text and puts it in the input box, so you can read and edit it before you send it.

Use it

  • Ctrl+Space (or /voice) starts recording. The input box shows a red ● rec timer and a level meter.
  • Ctrl+Space or Enter stops and transcribes. Esc cancels and throws the recording away.
  • A clip can be up to 2 minutes long. Recording stops by itself at the limit.
  • The text is inserted at the cursor and is never sent by itself.
  • Languages: Arabic and English, detected automatically. /voice lang ar or /voice lang en fixes the language, and /voice lang auto goes back to detection.
  • Line passes file names from your project as vocabulary hints, so code words come out right more often.
  • /voice off turns voice off again.

Terminals don't report when a key is released, so voice works as tap to start and tap to stop rather than hold to talk.

Privacy

Voice is off until you accept a one-time notice the first time you press Ctrl+Space. When you record:

  • The audio is sent to the Line service over HTTPS for transcription, and only the text comes back.
  • Line doesn't store your audio. The service keeps the clip length for metering and limits.
  • Nothing is recorded unless the red ● rec indicator is showing.

Limits

Per account: up to 20 clips a minute, 2 minutes per clip, and a daily total (60 minutes by default) that resets at 00:00 UTC. A clip that fails to transcribe doesn't count against the daily total.

Microphone setup

SystemHow Line records
WindowsThe default input device (Windows audio). Allow microphone access for desktop apps in Settings › Privacy › Microphone
macOSThe default input device. The first time, macOS asks whether your terminal app may use the microphone
LinuxThe first recorder found on your PATH: parecord (PulseAudio or PipeWire), pw-record, arecord (ALSA) or ffmpeg. Most desktops already have one; otherwise install pulseaudio-utils or pipewire
WSLWorks through WSLg's PulseAudio (parecord)

If recording can't start, Line says what is missing (for example, no recorder installed on Linux).