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

# Voice model and languages

> Choose the voice model Pipit uses and the languages it understands.

The voice model is what turns your speech into text. It runs on your Mac, so your audio never leaves it. You'll find it in **Home** > **Voice Model**.

## Choose English or Multilingual

Click the current voice model in **Home** > **Voice Model** to see the choices. Each one shows its name, a short description, and its size.

| Voice model | Description in Pipit | Choose it if |
| - | - | - |
| **English** | Most accurate | You dictate in English. This is the default. |
| **Multilingual** | 25 languages, less accurate for English | You dictate in a language other than English, or switch between languages. |

Click a voice model to switch to it. If it isn't on your Mac yet, Pipit downloads it first; a download arrow next to its size tells you it will.

## Languages

During setup, Pipit asks you to choose your language:

* **English Only** (marked RECOMMENDED): Pipit uses the **English** voice model.
* **Multiple Languages** (marked FLEXIBLE): detects and transcribes 25 European languages. Pipit uses the **Multilingual** voice model.

After setup, your language goes with the voice model you pick. To change languages, pick **Multilingual** or **English** in **Home** > **Voice Model**. **Reset Onboarding**, at the bottom of **Home**, also takes you back through setup, including the language choice.

**Multilingual** understands these 25 languages:

English, German, Spanish, French, Italian, Portuguese, Dutch, Polish, Russian, Ukrainian, Czech, Slovak, Hungarian, Romanian, Bulgarian, Croatian, Slovenian, Danish, Swedish, Finnish, Estonian, Latvian, Lithuanian, Greek, and Maltese.

You don't need to tell Pipit which one you're speaking.

## The download

Setup downloads what Pipit needs once: up to about 1.2 GB. You need an internet connection for the download. If you're offline, Pipit says "Connect to the internet to download the voice model."

Once a voice model is downloaded, Pipit works fully offline.

## Status and actions

The line under the voice model's name tells you where it is:

| Status | What it means |
| - | - |
| **Loaded** | Ready. Your next dictation starts right away. |
| **On disk** | Downloaded, but not in memory. Pipit loads it when you next dictate. |
| **Not downloaded** | Not on your Mac yet. |
| **Downloading** | Downloading, with a percentage. |
| **Warming up…** | Getting ready to use. |
| **Switching model…** | Changing to the voice model you picked. |
| **Offline** | Pipit needs the internet to finish the download. |

The button beside it changes with the status:

* **Download**: downloads the voice model and loads it.
* **Load**: loads a downloaded voice model into memory.
* **Unload**: frees the memory the voice model uses. It reloads when you next dictate.
* **Delete**: appears when the voice model is on disk but not loaded. Deletes its downloaded files to free disk space. Click **Delete**, then **Confirm**. You'll need to download it again to dictate with it.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.