Create speech with Text to Speech
Turn a script into natural speech, with a choice of voices, models and delivery settings.
Updated November 10, 2026 · 2 min read
On this page
Create a take
- Open Text to Speech.
- Type or paste your script. The counter shows how much of your plan's limit per generation you've used.
- Choose a voice. You'll see voices in use, your recent voices and More voices. On a desktop you can also browse the full list.
- Choose a model.
- Click Generate speech. The button shows what it will cost.
Choose a model
| Model | Best for |
|---|---|
| Vega V2 | The most expressive delivery, with style tags, in 30 languages |
| Vega V1 | Natural, conversational delivery |
| V1 (Classic) | Steady, studio-quality narration |
| V2 (Multi-lingual) | The same voices in more languages |
| V3 Turbo | The quickest results, in English only, with sounds such as [laugh] and [sigh] |
A lock icon means the model comes with a higher plan. You can compare plans to see which models each one includes.
If some voices are missing, the model you picked may not speak their language. The voice list tells you how many are hidden, and they come back when you switch model.
Adjust the delivery
Open the settings panel to fine-tune your take:
- Delivery: Choose Steady, Natural or Expressive, or set a custom mix.
- Pace: Slow down or speed up, from 0.5× to 2×.
- Expressiveness and Guidance: These apply to the Classic models.
- Variation: This applies to the Classic models and Vega V1.
- Format: Choose MP3 or WAV. Vega models always deliver WAV.
- Folder and Title: Keep your work organised in My Speeches. A title can be up to 50 characters.
Vega V2 uses Pace from these settings, and takes the rest of its direction from style tags in your script. Reset delivery returns everything to the defaults.
Add pauses and emotion
Type [ in the editor to see the tags your model supports, or use the Tags row above the script. To add a two-second pause, type [break 2]. See Add pauses and style with tags.
What it costs
The button shows the credits your script will use, and the footer shows your remaining balance. Credits are only charged when the audio is ready. If a take fails, your credits are returned. See How credits work.
While you wait
Takes show In the queue and then Generating with a percentage. You can keep working and queue more takes, up to your plan's limit. If you reach it, wait for one to finish.
Your takes are also saved in My Speeches, so nothing is lost if you leave the page.
Scripts that need review
Most scripts generate straight away. Occasionally one is held back before it's generated, and we'll email you when the review is done. Scripts that go against our usage policy aren't generated.