# Audio models

The Lock-in agent can make speech, music, sound effects and video soundtracks with 8 audio models. See what each does, its limits and what it costs in credits.

Audio models are available only inside the Lock-in agent, the **Chat** in Studio. There is no standalone speech, music or sound page. You describe what you want, and the agent proposes a generation with its price.

## The 8 audio models

| Model | Maker | Type | What it does | Limits | Credits |
| --- | --- | --- | --- | --- | --- |
| ElevenLabs v3 | ElevenLabs | Speech | Turns text into spoken audio. | Up to 5,000 characters per request | 184 per 1,000 characters |
| ElevenLabs Turbo v2.5 | ElevenLabs | Speech | Turns text into spoken audio. | Up to 5,000 characters per request | 92 per 1,000 characters |
| MiniMax Speech 2.8 HD | MiniMax | Speech | Turns text into spoken audio. | Up to 10,000 characters per request | 184 per 1,000 characters |
| MiniMax Speech 2.8 Turbo | MiniMax | Speech | Turns text into spoken audio. | Up to 10,000 characters per request | 111 per 1,000 characters |
| ElevenLabs Music v2.5 | ElevenLabs | Music | Makes music from a description; it can be instrumental. | 3 to 180 seconds | 1,103 per minute (30 seconds = 552) |
| Google Lyria 3.5 | Google | Music | Makes a song from a description. | The model picks the length | 184 per song |
| ElevenLabs Sound Effects v2 | ElevenLabs | Sound effects | Makes a sound effect from a description. | 0.5 to 22 seconds | 4 per second (5 seconds = 19) |
| MMAudio v2 | MMAudio | Video to audio | Adds sound to a video from the chat. | 1 to 30 seconds; needs a video from the chat | 2 per second (8 seconds = 15) |

## How audio is priced

Each model charges per unit: per 1,000 characters of text for speech, per minute for ElevenLabs Music v2.5, per second for sound effects and MMAudio v2, and per song for Google Lyria 3.5. The agent works out the price of your request from the characters or seconds it asks for and rounds it up to a whole credit.

In Ask mode the agent shows the exact price on an approval card before it runs. In Auto mode it runs without approval. See [Model pricing](https://www.uova.com/help/billing/model-pricing) for the rest of the price list.

## Use audio in the Lock-in agent

1. **Open Chat** In Studio, open **Chat**, the Lock-in agent. You need to be signed in.
2. **Describe the audio** Say what you need: a voice-over with its script, a song, a sound effect or a soundtrack for a video you made in the chat.
3. **Approve the proposal** The agent proposes a generation with its price. In Ask mode, check the price on the approval card and approve it.

## Limits

- Audio works only inside the Lock-in agent. You cannot open an audio model from **Create Image** or **Create Video**.
- Speech has a character limit per request: 5,000 characters on the two ElevenLabs speech models and 10,000 on the two MiniMax models.
- MMAudio v2 needs a video from the chat and makes 1 to 30 seconds of sound for it.
- UOVA does not offer voice cloning or lip-sync.

## Common questions

**Is there an audio page in Studio?**

No. Audio models run only inside the Lock-in agent. See [The Lock-in agent](https://www.uova.com/help/studio/lock-in-agent) for how the chat works.

**How much does a voice-over cost?**

It depends on the model and the length of the script. ElevenLabs Turbo v2.5 costs 92 credits per 1,000 characters, and ElevenLabs v3 and MiniMax Speech 2.8 HD cost 184. The approval card shows the exact price of your request.

**Can the agent make a song of a set length?**

ElevenLabs Music v2.5 makes 3 to 180 seconds, and a 30-second piece costs 552 credits. Google Lyria 3.5 picks the length itself and costs 184 credits per song.

Source: https://www.uova.com/help/models/audio