Skip to content
Voice model · ElevenLabs

ElevenLabs v3 when one sentence should be warmer than the next

The other engines set the mood across the whole read — with one slider or one emotion from a list. This one reads tags written into the copy, so the opening line can be calm and the line about plastic-free packing can be delighted. What you give up for that is everything else: no speed, no style, no similarity. The square brackets are the whole of the control.

Rendered by this model

Turn the sound on and listen

–:–– / –:––

The script the engine was given

[warmly] Nowa kolekcja jest już dostępna. [pause] Bawełna, len i jeden kolor, który schodzi pierwszy. [excited] A całe zamówienie pakujemy w jeden karton, bez grama folii.

Read in Polish by Sarah. The script carries three tags — [warmly], [pause] and [excited] — and none of them was read out loud; speech recognition returned the advert copy alone, which is the only way to tell a tag that was performed from one that was ignored.This was generated by the model this page is about, exactly the way the app runs it. It is not a photo shoot and not client work.

Specification

Built by
ElevenLabs
Speaks Polish
yes — checked on a Polish sentence
Ready-made voices
20
Steering the delivery
tags written into the script
Character limit
5,000
File from the engine
mp3
Render time
about 15s
Cost
1 credit per 1000 characters, rounded up

What it is good for

The tone changes on the sentence you choose
A tag takes effect from where it stands, so an advert can open calmly and lift on the offer. No other engine on this list lets the mood change inside one recording.
Nothing to dial in
"[warmly]" is understood on the first try, unlike the question of whether stability should be 0.4 or 0.55. The script reads like a screenplay and is edited like one.
The same twenty voices as the other tiers
The timbre does not change when you move between the three ElevenLabs engines, because the list of names is shared. All twenty were confirmed by submitting each one as a job.

When to pick something else

This is the most constrained of the three ElevenLabs tiers in terms of fields: no speed, no style, no similarity, and a 5000-character limit. If an advert has to fit a fifteen-second slot, the speed cannot be raised here — ElevenLabs Turbo v2.5 is for that. If you are recording forty descriptions in one voice, repeatability is easier to set with the sliders on ElevenLabs Multilingual v2, which also expands abbreviations and clock times for you. And if you need silence of a stated length rather than a mood, the pause markers are on MiniMax Speech 2.8 HD.

See every model

How to brief this model

The things this engine handles differently from the rest. The habits that hold for every model are collected separately, in the help centre.

  1. Put the tag in front of the sentence it should affect, not at the top of the script. It takes effect from where it stands.
  2. Listen to the opening after a render. An engine that does not understand a tag reads it out — ours performed [warmly], [pause] and [excited], but that is the one thing to check with a tag you have not used before.
  3. Do not write speed in as a word. This tier has no such field, and "[faster]" will be read as a mood or ignored — use Turbo v2.5 to fit a duration.
  4. Stay inside 5000 characters per job. That is a lower limit than MiniMax, so a long product description has to be split here.
  5. Do not mix tags inside one sentence. Two moods on one sentence gives you a recording with a seam in it rather than a change of tone.
How to write a brief that works

Questions about this model

Try this model on your own product

Sign up without a card and get 40 credits. Switching model is one click, so you can send the same brief through several of them and keep the best frame.