# Voice Lab


# Voice Lab

Voice Lab is where you choose how your agents sound. Clone a custom voice from a short audio sample, or pick from a library of ready-made voices — then preview it and assign it to any agent. The right voice makes your calls feel consistent and on-brand, and the same voice can back as many agents as you like.

Open **Voice Lab** (`/voice-lab`).

## Two ways to get a voice

![Voice Lab](/img/screenshots/portal/voice-lab.png)
*Voice Lab: the Quick Listen playground and the prebuilt voice library (Explore, Default Voices, My Voices).*

You have two paths, and most organizations use a mix of both:

- **Clone a custom voice** — capture a specific person's voice (a brand spokesperson, a founder, a named agent persona) from a short recording.
- **Pick a prebuilt voice** — choose from a ready-made library when you just need a good, professional voice quickly and don't need it to be anyone in particular.

Whichever you pick, you preview it in Voice Lab first, then assign it to one or more agents.

## Clone a custom voice

Instant cloning creates a custom voice from a short recording — a sample of **at least 10 seconds** is enough to get started.

1. Provide a clean voice sample (10 seconds or more).
2. Create the clone.
3. **Preview** it to hear how it sounds before you use it.

:::tip

Quality in, quality out. Record the sample in a quiet room with a single speaker and no background music. Have the speaker talk naturally, the way they would on a real call. A clean 10–30 second sample clones far better than a long, noisy one.

:::

:::warning

Only clone a voice you have the right to use. Get explicit consent from the person whose voice you're cloning — this is both an ethical and a legal requirement in many places.

:::

## Prebuilt voice library

If you don't need a custom voice, choose from the **prebuilt voice library** — a selection of ready-made voices. Browse the options and **preview** any of them to hear it before deciding. Listen with your real use case in mind: a calm, measured voice suits support and confirmations, while a brighter, warmer voice can suit outreach.

## Assign a voice to agents

Once you've cloned or chosen a voice, **assign it to your agents**. From then on, that agent speaks in the assigned voice on every call. Because a voice is independent of any one agent, you can give a whole team of agents the same on-brand voice, or give each agent its own.

To change how an agent sounds later, assign a different voice here, or fine-tune the delivery on the agent (see below).

## Shape the delivery on the agent

Voice Lab decides *which* voice an agent uses. *How* that voice is delivered — stability, style, speed, pitch, emotion and pronunciations — lives on the agent itself, on its [Voice & audio](./agents/voice-audio) tab. Pick the voice here, then shape its pacing and expressiveness there.

## Next steps

- [Tune voice & audio](./agents/voice-audio) on an agent to shape delivery.
- [Build an agent](./agents/create-agent) and give it your chosen voice.