Question 1

What can I make with ElevenLabs Text to Speech?

Accepted Answer

Any spoken audio that starts as a script. Generate a voiceover for a YouTube video or a TikTok. Narrate a product demo or an explainer. Read an ad in three different voices and keep the best take. Voice a chapter of an audiobook, a training module, or a line of game dialogue. Each run returns an MP3 you can drop into Premiere Pro, DaVinci Resolve, CapCut, Final Cut Pro, or any editor that accepts MP3.

Question 2

Should I pick Eleven v4, Eleven v3, Multilingual v2, or Flash v2.5?

Accepted Answer

Start with Eleven v4. It is the newest and most natural of the four, it follows audio tags, it speaks 85 languages, and it reads up to 10,000 characters in one run. Keep Eleven v3 for a script you already tuned on it. Pick Multilingual v2 for long narration where a steady, even read matters more than emotion, like an audiobook chapter or a course. Pick Flash v2.5 when the script is long or you need many takes, since it costs half as much and reads up to 40,000 characters in one run. All four use the same voices, so you can switch models and keep the voice.

Question 3

How do audio tags like [whispers] and [laughs] work?

Accepted Answer

Write a direction in square brackets where it should happen, like [whispers], [laughs], [sighs], [excited], or [sarcastic]. Eleven v4 and Eleven v3 act it out instead of reading it aloud. You can use several tags in one script, or even in one sentence. The voice still matters, so a calm narrator will not suddenly shout on cue. Only Eleven v4 and Eleven v3 follow tags, so remove them before you switch to Multilingual v2 or Flash v2.5.

Question 4

Which voices can I use, and can I clone my own voice?

Accepted Answer

Pick from 21 ElevenLabs stock voices, with American, British, and Australian accents. George, a British storyteller, is the default. Brian suits deep, calm narration, Jessica a playful read, Daniel a broadcast tone, and Liam social media content. For a voice of your own, describe it in ElevenLabs Voice Design, pick one of the three samples, and add it here as the designed voice. Voice cloning is not available, so a recording of a real voice cannot be used.

Question 5

Which languages does it speak?

Accepted Answer

Eleven v4 speaks 85 languages, Eleven v3 speaks 74, Flash v2.5 speaks 32, and Multilingual v2 speaks 29. Write the script in the language you want to hear, and the voice reads it in that language. Every stock voice can speak every supported language, though its native accent can carry over. Try two or three voices for a new language and keep the one that sounds most natural.

Question 6

How long can my text be?

Accepted Answer

Eleven v3 reads up to 5,000 characters per run, Eleven v4 and Multilingual v2 up to 10,000, and Flash v2.5 up to 40,000. As a rough guide, 1,000 characters make about a minute of speech. For a longer piece, like a full audiobook chapter, split the script at paragraph breaks and generate each part. If the text is too long for the model you picked, Visual Sandbox tells you before it charges anything.

Question 7

How do I fix a word it pronounces wrong?

Accepted Answer

Spell it the way it sounds. A name like Siobhan reads better as Shiv-awn. Write numbers, dates, and abbreviations as words, especially with Flash v2.5, which reads them less reliably than the other models. Use punctuation for rhythm. A comma adds a short pause, and an ellipsis adds a longer one. A single line costs very little, so test a tricky line on its own before you voice the whole script.

Question 8

How is it different from the other ElevenLabs models?

Accepted Answer

Text to Speech turns written text into a spoken voice. Voice Design creates a new voice from a written description, for Text to Speech to speak with. Voice Changer takes a recording you already have and speaks it again in another voice, with the words, the timing, and the emotion kept. Scribe does the reverse and turns the speech in an audio or video file into a transcript. Sound FX makes audio that is not speech, like footsteps, whooshes, and ambiences, and Eleven Music makes full songs. A typical project uses several of them: a voiceover from Text to Speech, a music bed from Eleven Music, and effects from Sound FX, all on one Visual Sandbox canvas.

Question 9

How much does ElevenLabs Text to Speech cost?

Accepted Answer

Pricing is billed per character of your text, so a one-line test costs a fraction of a full script. Spaces and audio tags count as characters. Eleven v4, Eleven v3, and Multilingual v2 share one rate, and Flash v2.5 costs half as much. Visual Sandbox shows the exact price before you generate. There is no subscription and no monthly character quota.

Question 10

Can I use the voiceovers commercially?

Accepted Answer

Yes. ElevenLabs allows commercial use of generated speech, including monetized YouTube videos, ads, podcasts, audiobooks, apps, and client work. You are responsible for the script and the final use. Some platforms ask you to label AI-generated voices, so check the rules where you publish.

Tier	Price
Eleven v4	$0.13 / 1K characters
Eleven v3	$0.13 / 1K characters
Multilingual v2	$0.13 / 1K characters
Flash v2.5	$0.07 / 1K characters

Create with ElevenLabs Text to Speech

What is ElevenLabs Text to Speech AI voice generator?

ElevenLabs Text to Speech pricing