Docs

Voice

Speak into the composer and Raven transcribes it.

The microphone button in the composer lets you speak instead of type. Your speech is transcribed into the composer, you edit it if you want, and you send it like any other message.

Using it#

1. Press the microphone button in the composer.
2. Speak. The recording indicator tracks your voice and settles when you stop talking.
3. The transcription lands in the composer. Read it over, edit if needed, then send.

How it behaves#

• Listening is local. The app decides when you are talking and when you have stopped with on-device voice detection, so the recording light follows you rather than a timer.
• Audio is bounded. The capture buffer holds a bounded tail of recent audio. A very long recording drops the oldest audio and tells you it did, instead of growing until the request fails.
• Failures retry in place. A transient transcription error retries automatically with backoff, and text you already have is never inserted twice.
• It stops when you leave. Switching tabs, locking the screen, or backgrounding the app stops the recording, so nothing captures while you are away.

Tips#

• Say punctuation out loud: "comma", "period", "new paragraph".
• Dictate in short turns and review before sending; a corrected sentence beats a misunderstood one.
• Voice works the same on the web app, on desktop, and on Android.