> ## Documentation Index
> Fetch the complete documentation index at: https://docs.parsaa.app/llms.txt
> Use this file to discover all available pages before exploring further.

# Voice Input

> Dictate messages to Parsaa instead of typing

## Overview

Parsaa can transcribe speech into the composer. Click the **microphone** button in the chat input bar, speak, and your words are transcribed and inserted as text you can edit before sending.

## How to Use

<Steps>
  <Step title="Click the mic">
    The microphone button ("Dictate message") sits in the composer toolbar. The first time, macOS asks for microphone permission.
  </Step>

  <Step title="Speak">
    The composer shows a live waveform and a timer while recording. Click stop when you're done, or cancel to discard.
  </Step>

  <Step title="Review the text">
    Parsaa transcribes the audio and drops the text into the composer. Edit it like any message before sending.
  </Step>
</Steps>

<Tip>
  If you have more than one microphone, use the menu next to the mic button to pick an input device. Your choice is remembered.
</Tip>

## Transcription Providers

Choose how audio is transcribed in **Settings → Preferences → Transcription**:

<CardGroup cols={2}>
  <Card title="Parsaa Cloud" icon="cloud">
    Audio is transcribed in the cloud. Requires an internet connection. This is the default.
  </Card>

  <Card title="Apple On-Device" icon="apple">
    Transcription runs locally with Apple's Speech framework — no network. Requires a one-time per-language model download.
  </Card>
</CardGroup>

<Note>
  With **Apple On-Device**, if the language model isn't installed yet, Parsaa temporarily falls back to Parsaa Cloud so you can keep working while the model downloads.
</Note>

## Permissions

Voice input needs microphone access. If it's denied, enable it in **System Settings → Privacy & Security → Microphone** and re-try.