# VoiceStudio

> Connect your own VoiceStudio so assistants can speak a script, design a synthetic voice and transcribe recordings. Voice cloning is never offered.

VoiceStudio is a free, open source voice program you run yourself. It turns text into speech, designs voices from a description, transcribes recordings and dubs in many languages. Once you add your own VoiceStudio as a server, your assistants can use it to narrate a chapter, make a voiceover or write captions.

StellarFirm does not host VoiceStudio and does not ship it. Your assistants only talk to the copy you run, over the internet.

## Which assistants use it

- [Copywriter](/docs/personas/copywriter), for audiobooks and narration
- [Marketing](/docs/personas/marketing), for voiceovers and captions

## What assistants can do

| Ability | How it runs |
| --- | --- |
| List voices, languages and personalities | Reads on its own |
| Describe a voice in words | Reads on its own |
| Speak a script with a voice | Runs on its own when you asked for it |
| Design a synthetic voice from a description | Runs on its own when you asked for it |
| Transcribe a recording | Runs on its own when you asked for it |
| Check that your VoiceStudio is up | Runs on its own |
| Copy or clone a voice | Never offered |

Nothing is published, posted or sent by this tool. Publishing a video or an audiobook still waits for your Approve, like everything else.

## Setup

1. Run VoiceStudio on your own computer or server with an access key set, so only you can use it.
2. Put it on a public https address, for example with a Cloudflare Tunnel or Tailscale Funnel. The address must be on the standard https port. A plain local address such as `127.0.0.1` cannot be reached from StellarFirm cloud.
3. Open [Integrations](https://stellarfirm.ai/app/integrations), choose **Add your own server**, and paste the https address of your VoiceStudio, ending in `/mcp/`.
4. Choose the bearer key option and paste the access key you set in step 1.
5. Pick the assistants that may use it and save.

## Good to know

- Voice cloning is never offered, for any server you connect. If a server lists a tool that clones a voice, StellarFirm hides it and the assistant cannot call it. Assistants never imitate a real or named person's voice.
- Each voice model comes with its own licence, and VoiceStudio shows it for every model. Check that it allows your use, such as selling an audiobook, before you publish.
- Label AI voices the way each platform or retailer asks.
- Your key stays with your StellarFirm workspace and is never shown back to you.

## Prompts to try

```prompt title="Chapter sample"
Copywriter, use my VoiceStudio to read the first page of chapter 1 aloud in a warm, unhurried voice. Make a short sample first.
```

```prompt title="Voiceover"
Marketing, design a calm, friendly synthetic voice and record this 30 second script for my product video: [script].
```

---

Source: https://stellarfirm.ai/docs/integrations/voicestudio