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

# Voice Input

> Speak instead of typing — dictate tasks, notes, comments and instructions to the AI agent.

Voice Input puts a microphone next to the places you write. Speak, and what you
said becomes text you can read and edit before anything is saved.

<Note>
  Voice Input is available on **Startup** plans and above, and is switched on by
  default. Turn it off for a workspace under **Settings → Modules**.
</Note>

## Where you'll find it

<CardGroup cols={2}>
  <Card title="AI agent" icon="sparkles">
    In the composer, next to the send hint. Speak an instruction and the agent
    treats it exactly like one you typed.
  </Card>

  <Card title="Descriptions" icon="align-left">
    Task and project descriptions, on both the create dialogs and the task
    side panel.
  </Card>

  <Card title="Comments and notes" icon="message-square">
    Task comments, project notes, personal notes, and the inbox.
  </Card>

  <Card title="Anywhere you write" icon="mic">
    The mic lives inside the shared editor, so any writing surface added later
    has it from the start.
  </Card>
</CardGroup>

## Using it

<Steps>
  <Step title="Press the microphone">
    The first time, your browser asks for microphone permission. Worksynk can't
    record until you allow it.
  </Step>

  <Step title="Speak">
    A live meter moves with your voice, so you can see it's hearing you. Pause
    and resume as often as you like — the recording continues where it stopped.
  </Step>

  <Step title="Stop, or discard">
    **✓** transcribes what you recorded. **✕** throws it away without sending
    anything.
  </Step>

  <Step title="Read it before you save">
    The text lands at your cursor, mixed into whatever you'd already written
    rather than replacing it. Nothing is submitted for you — edit it, then save
    as normal.
  </Step>
</Steps>

## Talking to the AI agent

The agent treats a spoken instruction exactly like a typed one, so anything it
can do, you can now ask for out loud:

> "Create a bug in the mobile app project, assign it to Karthik, mark it high
> priority, and make the deadline next Monday."

The agent works out what you meant and shows you the action it's about to take —
project, work type, assignee, priority, due date — and **waits for you to
confirm**. Nothing is created until you press confirm.

<Warning>
  Transcription is very good but not perfect, especially with names and
  technical terms. Read what came back before you save or confirm it.
</Warning>

## Privacy

* **Recordings are never stored.** Audio is sent for transcription and
  discarded with the request. There is no audio file to find, download, or
  leak.
* **Transcribed text is ordinary text.** It lives wherever you put it — a task,
  a note, a comment — under exactly the permissions that place already has.
  Voice creates no new way to see anything.
* **Recording only happens while you hold the mic open.** Your browser shows
  its own recording indicator throughout, and Worksynk releases the microphone
  as soon as you stop.

## Credits

Each transcription costs **1 AI credit**, the cheapest capability in the
product. Usage appears under **Settings → AI** alongside everything else, and
workspaces using their own API key are not charged credits at all.

## Troubleshooting

<AccordionGroup>
  <Accordion title="I can't see the microphone anywhere">
    Three things hide it: the **Voice Input** module being off for your
    workspace (Settings → Modules), a plan below Startup, or a browser that
    can't record. Firefox has no support for the recording API Worksynk uses —
    Chrome, Edge and Safari all work.
  </Accordion>

  <Accordion title="It says microphone access is blocked">
    The browser is refusing, not Worksynk. Allow microphone access for this
    site in your browser's site settings, reload, and try again. On macOS also
    check System Settings → Privacy & Security → Microphone.
  </Accordion>

  <Accordion title="Nothing was picked up">
    The recording captured silence. Check the right input device is selected in
    your system sound settings, and that nothing else is holding the microphone
    — video call apps commonly do.
  </Accordion>

  <Accordion title="The transcription failed">
    Usually a dropped connection. Try again; if it keeps failing, type instead
    and let your workspace admin know so they can check AI credits under
    Settings → AI.
  </Accordion>

  <Accordion title="Long recordings">
    A single recording can run up to ten minutes. For anything longer, record
    it in sections — each one lands at your cursor, so they join up in order.
  </Accordion>
</AccordionGroup>