---
title: Audio Studio
description: "Text-to-speech with one voice or a dialogue, and music from a description — Clipia Audio Studio: models, settings, prices in credits and tips."
icon: AudioLines
---

Audio Studio ([/create-audio](/en/create-audio)) is the workspace for sound: voice-over for a video, a dialogue between several voices and music from a text description. Each mode is a separate tab, and the exact price is shown on the start button before any credits are charged.

<Callout type="info" title="Modes are rolling out gradually">
Transcription, speech translation, recording cleanup and personal voices are still in development.
</Callout>

## Studio modes

<ModeList>
  <Mode icon="Speech" title="Voiceover" models="Gemini 3.1 Flash TTS, Gemini 2.5 Pro TTS, ElevenLabs v4, ElevenLabs v4 Turbo" price={<>Gemini 3.1 Flash TTS — 4 credits per 1,000 characters</>} href="#voiceover">
    A voice track from text: a single narrator or a dialogue between several voices. Russian and English.
  </Mode>
  <Mode icon="Music" title="Music" models="Suno Music (Suno)" price={<>12 credits per run</>} href="#music">
    A background track or a song from a description: instrumental or with vocals.
  </Mode>
</ModeList>

The studio does not generate sound effects (footsteps, explosions, ambient noise): its models produce speech and music.

## Tab addresses

Every available studio tab has its own address, so you can bookmark it or send it to a colleague.

| Tab | Address |
| --- | --- |
| Voiceover (opens by default) | [/create-audio](/en/create-audio) |
| Music | [/create-audio/ai-song-generator](/en/create-audio/ai-song-generator) |

In the studio, dialogues can use more than two voices, and every job appears in **My tasks** with its status.

## How credits are charged

<Steps>
<Step>
### Price before you start

The studio asks the server for a quote and shows the total on the start button. A quote is valid for 10 minutes and refreshes automatically. If the rate changes in the meantime, you will see the new price before anything is charged.
</Step>
<Step>
### Hold on start

When you start a job, the credits are put on hold: the studio header shows "N credits awaiting confirmation". Pressing the same button again does not create a second job or charge you twice.
</Step>
<Step>
### Charge or refund

When the result is ready, the hold becomes a charge ("N credits charged"). If the job never reached the model (it was rejected before being sent), the credits are returned to your balance.
</Step>
</Steps>

If your balance is too low, the job does not start and the studio shows the price next to your current balance. To top up, see [Credits and plans](/en/docs/product/credits).

**How the price is calculated:**

- **Text** is billed by character count, rounded up to a whole credit: the model price is per 1,000 characters, so 500 characters cost half of it and 10,000 characters cost ten times as much. The minimum is one credit.
- **Music** has a fixed price per run.

<PriceTable
  caption="Audio Studio rates"
  rows={[
    { name: 'Voiceover — ElevenLabs v4', credits: 7, note: 'per 1,000 characters' },
    { name: 'Voiceover — ElevenLabs v4 Turbo', credits: 4, note: 'per 1,000 characters' },
    { name: 'Voiceover — Gemini 3.1 Flash TTS', credits: 4, note: 'per 1,000 characters' },
    { name: 'Voiceover — Gemini 2.5 Pro TTS', credits: 4, note: 'per 1,000 characters' },
    { name: 'Music — Suno Music', credits: 12, note: 'per run, all returned variants' },
  ]}
/>

<Callout type="info" title="Prices">
Prices on this page come from the service pricing and are refreshed together with the documentation. The exact cost is always shown on the start button before you are charged.
</Callout>

These actions cost no credits: listening to voice samples, translating a prompt into English and dictating a prompt by voice. New here? Try it free — 5 credits on sign-up.

## Voiceover

A voice track for a video, lesson, presentation or ad. Use it for voice-over, short dialogues and narrating parts of a book.

<Steps>
<Step>
### Enter text and pick a voice

Paste your script, choose a model and a voice. You can listen to voice samples before starting.
</Step>
<Step>
### Start

Press **Create voiceover**. The price is shown on the button.
</Step>
<Step>
### Listen and download

The finished track opens in the result area. Download the MP3; the job is saved to your history.
</Step>
</Steps>

### Models

| Model | Developer | Voices | Text per request | Price |
| --- | --- | --- | --- | --- |
| Gemini 3.1 Flash TTS | Google | 30 (16 male, 14 female) | 10–500 characters | 4 credits / 1,000 characters |
| Gemini 2.5 Pro TTS | Google | 30 | 10–500 characters | 4 credits / 1,000 characters |
| ElevenLabs v4 | ElevenLabs | 21 | 10–10,000 characters | 7 credits / 1,000 characters |
| ElevenLabs v4 Turbo | ElevenLabs | 21 | 10–10,000 characters | 4 credits / 1,000 characters |

Gemini 3.1 Flash TTS is selected by default in the studio. For help choosing, see [Audio and speech models](/en/docs/models/audio).

### Settings

<Tabs items={['Gemini', 'ElevenLabs']}>
<Tab value="Gemini">

Gemini models let you direct the narrator:

- **Delivery style** — empathetic, vocal smile, newscaster, promo, deadpan.
- **Pace** — natural, rapid fire, staccato.
- **Accent** — neutral or one of the English accents: American, American Valley, American South, British RP, British Brixton, Transatlantic, Australian.
- **Variability** — higher values give a more expressive, less predictable read; lower values are more stable.
- **Narrator profile** — timbre, age, character and manner.
- **Scene** — setting and emotional context.

</Tab>
<Tab value="ElevenLabs">

- **Delivery** — Expressive, Balanced or Even.
- **Voice similarity** — how closely the model sticks to the timbre of the chosen voice.
- **Seed** — a number that helps reproduce a similar read with the same settings.
- **Previous text** — up to 100 characters of what was said before this fragment, to keep intonation consistent across parts.
- **Delivery tags** — written right in the text: `[whispering]`, `[laughing]`, `[excited]`, `[sad]`.

</Tab>
</Tabs>

### Dialogue

Dialogue is available in ElevenLabs v4 and v4 Turbo. Press **Add voice**, pick a voice for each role and start each line with its number: `1:`, `2:` and so on, one line per row. You can include up to 20 lines per run. Voice labels are not billed — only spoken characters count.

### Language

The speech language is detected from the text: if it contains Cyrillic, it is read in Russian, otherwise in English. Russian and English are supported; text in other languages will be read as English or Russian.

### Tips

- Write numbers, dates and abbreviations the way they should sound: "May twenty-fifth" rather than "05/25".
- Punctuation controls pauses: a comma gives a short pause, a full stop or new paragraph a longer one.
- Split a long script into parts of up to 10,000 characters (ElevenLabs) or 500 characters (Gemini) and voice them one by one.

## Music

A background track for a video, intro or presentation, or a song with vocals, from a text description. The model is Suno Music by Suno. It costs 12 credits per run, and the price covers every variant the model returns.

<Steps>
<Step>
### Describe the track

Specify genre, mood, tempo, instruments and the scene the music is for.
</Step>
<Step>
### Choose the arrangement

**Instrumental** for a backing track without voice, **With vocals** for a song.
</Step>
<Step>
### Listen and download

Press **Create music**. The model chooses the track length. For another variant, change the description and run again — each run is billed separately.
</Step>
</Steps>

Avoid contradictory wishes in one description ("slow and very fast"): the model will pick one of them unpredictably.

## My tasks: history and statuses

Below the result area there is an expandable **My tasks** list. Each row shows the mode, date and time, and status:

| Status | Meaning |
| --- | --- |
| Queued | The job is waiting to be sent to the model |
| Sending, in progress | The model is processing the job |
| Done | The result is ready to play and download |
| Failed | The job was not completed |
| Unconfirmed | The result was not confirmed; credits stay on hold with no automatic refund |

**No cancellation.** A started audio studio job can't be cancelled: once sent, the model is already working. If a job never reached the model, the credits are returned automatically.

**Downloading.** The **Download** button lets you choose a format. If conversion to that format is unavailable or the recording is too large, download the original.

Finished tracks also appear in your general history — see [My works and tasks](/en/docs/product/my-works).

<Callout type="idea" title="Save your results">
Download finished tracks as soon as they are ready: the studio task list is tied to your browser and may not show up on another device.
</Callout>

## Troubleshooting

<Accordions>
<Accordion title="“The model declined the request under its content rules”">
The text or description breaks the model's rules. Rephrase the passage or remove the problematic content. See [Content rules](/en/docs/product/rules).
</Accordion>
<Accordion title="“The model service is temporarily unavailable”">
The model is not responding. Try again in a few minutes or choose another speech model.
</Accordion>
<Accordion title="“The model returned a flawed take and we rejected it”">
The studio checks results and does not deliver broken tracks. Change the text or voice and run again.
</Accordion>
<Accordion title="“The track is taking longer than usual”">
The job is not lost — check the generation queue or My tasks in a few minutes.
</Accordion>
<Accordion title="“Not enough credits”">
The job did not start and no credits were charged. Top up your balance and run it again.
</Accordion>
<Accordion title="Credits are “awaiting confirmation” and not returned">
If a job's result was not confirmed, the credits stay on hold: there is no automatic refund and such jobs are reviewed manually. Contact support with the mode, date and time of the job from My tasks.
</Accordion>
</Accordions>

## See also

<FeatureGrid cols={3}>
  <Feature icon="AudioWaveform" title="Audio and speech models" href="/en/docs/models/audio">Compare models for speech and music.</Feature>
  <Feature icon="Coins" title="Credits and plans" href="/en/docs/product/credits">How to top up and how refunds work.</Feature>
  <Feature icon="Video" title="Video Studio" href="/en/docs/studios/video">Add voice-over and music to your video.</Feature>
</FeatureGrid>