# Pictures and sounds

Add your company's pictures and MP3 sounds in Works Studio. Kapi keeps them for the company, places them where you say, and writes a sound's captions from what it hears.

A course names its pictures and sounds in `kapi-works.json`. Kapi keeps their files for your company, apart from the course, so a bundle carries no picture or sound bytes.

## Add them in Works Studio

On a course's page in Works Studio, the **+** in the chat adds pictures and sounds:

* Pictures: PNG, JPEG or WebP, up to 6 MB each.
* Sounds: MP3, up to 6 MB each, such as a welcome from your manager or a recorded call.
* Up to 10 with one change. Say where each one goes, and Kapi places it.

In a video, each lesson starts with its film, so Kapi places a sound after the film.

Each file goes straight to Kapi's storage. Kapi names it by its SHA-256, not by its file name, and keeps it while a course uses it.

## A sound in the course

`kapi-works.json` lists the sound by its SHA-256:

```json
{
  "id": "welcome-sound",
  "kind": "audio",
  "file": "media/welcome-sound.mp3",
  "sha256": "sha256:…",
  "speech": true,
  "captions": { "en-GB": "media/welcome-sound.en-GB.vtt" }
}
```

A page plays it with an audio block:

```json
{ "type": "audio", "media": "welcome-sound" }
```

## Captions

A sound with speech needs captions: a WebVTT file under `media/` for each language. The course carries its captions as text. Kapi checks this: [`captions_missing`](https://docs.kapiworks.com/build/reference/checks/captions-missing).

When you add a sound in the chat, Kapi listens to it with Gemini and writes its captions, one sentence at a time with its times. Music and sound effects need none.

## What learners get

* The sound plays with the player's own controls, which keyboards and screen readers know.
* The words being said show under it, and its whole transcript opens below it, for anyone who is deaf or has the sound off.

## Cost

Listening to a sound uses your company's AI credits, with the change that adds it.