Tools

ElevenLabs alternatives: cloud voices and free local ones

Nemo · · 7 min read

The main ElevenLabs alternatives fall into two groups: cloud voices billed per character, such as OpenAI’s speech API and Amazon Polly, and open voice models you run on your own computer, such as Kokoro and Piper, which have no per character cost. ElevenLabs is still the most expressive option for many voices. The alternatives win on price, privacy or working offline.

What is ElevenLabs and how does it charge?

ElevenLabs is a cloud voice service. You send text and its servers return speech. It also offers voice cloning, sound effects and music, and its text to speech models cover from 29 to more than 90 languages depending on the model.

It charges in credits. These plans are from the ElevenLabs pricing page, checked in November 2026:

Plan Price per month Credits per month
Free $0 10,000
Starter $6 30,000
Creator $22 121,000
Pro $99 600,000

Text to speech costs 1 credit per character on the Multilingual v2 model, and every ElevenLabs product draws from the same credit pool. The page lists a commercial licence from the Starter plan up. Larger plans (Scale, Business, Enterprise) exist for teams.

Do the arithmetic and Creator works out to roughly $180 per million characters, and Pro to about $165. A 60 second script is around 150 words, or roughly 900 characters, so Creator covers about 130 such narrations a month if you never regenerate a line. Retakes count: every new take spends credits again.

Why do people look for an ElevenLabs alternative?

  • Cost per character. The price buys real quality, but it is a meter. Daily publishing, long scripts and retakes all push you up a plan.
  • Wanting offline. A cloud voice needs a connection and sends your script to a server. Some creators cannot upload their text, or would rather not.
  • Needing less. If you want one clear narrator in English, you may be paying for cloning and a voice library you never open.
  • A shared pool. Sound effects and music spend the same credits as narration.

ElevenLabs alternatives compared

Option What it makes Where it runs How it charges Who it suits
ElevenLabs Expressive speech, voice cloning, sound effects, music Cloud Monthly credits Creators who want the most natural delivery
OpenAI speech API Speech from text Cloud, through an API Pay as you go, per character or token Developers and apps that call an API
Amazon Polly Speech from text, several voice engines Cloud, through an API Per character, with a free tier High volume at a low, predictable rate
Google Cloud Text-to-Speech Speech from text Cloud, through an API Per character, by voice type Teams already on Google Cloud
Kokoro Speech from text, fixed voices Your own computer Free, open (Apache licence) Offline narration in English
Piper Speech from text, downloadable voices Your own computer Free, open (GPL-3.0 licence) Local speech from a command line or script
Chatterbox Speech with voice cloning from a sample Your own computer Free, open (MIT licence) Tinkerers who accept an unmaintained project

What do the other cloud voices cost?

OpenAI lists its tts-1 model at $15 and tts-1-hd at $30 per million characters on its API pricing page. Amazon Polly lists Standard voices at $4, Neural at $16 and Generative at $30 per million characters, with a free tier that includes 5 million Standard characters a month. Google Cloud Text-to-Speech also bills per character, at rates that depend on the voice type. We could not read its price table when we checked, so see the page for current numbers.

On paper that is several times cheaper than ElevenLabs. Two cautions. These are services for developers: there is no studio where you paste a script and pick a take, unless an app you use connects to them. And price is not quality. ElevenLabs voices usually carry more emotion and variety, which matters for storytelling.

Are there free ElevenLabs alternatives?

Yes, in three forms.

  1. The ElevenLabs free plan. 10,000 credits a month is roughly ten minutes of speech. The commercial licence is listed from Starter, so read the terms before you publish monetized work made on the free plan.
  2. Free tiers of cloud services. Polly’s free tier is generous for Standard voices and smaller for the better engines.
  3. Open local models. No meter at all. You pay with setup time and your computer’s effort.

The local models differ:

  • Kokoro is a small model (82 million parameters) with Apache-licensed weights and 54 voices across 8 languages. Most of those voices are English. See what Kokoro TTS is and how to run it for the detail.
  • Piper describes itself as a fast, local text to speech engine, with a command line tool, a web server and a Python API. It is GPL-3.0 licensed.
  • Chatterbox, from Resemble AI, can clone a voice from a reference clip and is MIT licensed. Its repository says that as of August 2026 it is retired research and no longer maintained, and that every output carries an inaudible watermark.

What do you give up with a local voice?

  • Expressiveness. The top cloud voices still act better: pauses, emphasis, emotion.
  • Choice. A local model has a fixed set of voices. ElevenLabs has a large library, plus cloning.
  • Languages. Small local models are at their best in English.
  • Convenience. Running a model yourself means Python, a terminal and some patience, unless an app wraps it for you.

What you get back: no per character cost, no upload, and narration that works without a connection. The wider trade is covered in local AI vs cloud AI for making videos, and the post on offline AI video generators shows where a local voice sits in a full video workflow.

Where Bitfaceless fits

Bitfaceless is not a voice tool. It is a desktop app that turns a script into a short narrated vertical video with a pixel avatar host, motion graphics and word by word captions. It belongs on this page because of how it handles the voice.

It has local English voices (US and UK, male and female), built on the open Kokoro model, that run offline on your computer. There is no per character charge for them. If you prefer ElevenLabs, you can connect your own API key and use the voices on your account, billed by ElevenLabs. So you can narrate everyday videos locally and keep ElevenLabs for the ones where delivery matters most.

The limits: the local voices are English only, and the app does not clone your voice. It makes videos, so if you only need audio for a podcast or an audiobook, use a voice tool from the table. It runs on Mac and Windows (Apple silicon on macOS 14 or later, or a 64-bit PC on Windows 10 or 11). It is paid once with a 7-day refund: see the price or download Bitfaceless.

Frequently asked questions

Is there a free alternative to ElevenLabs?

Yes. Open models such as Kokoro and Piper are free to run on your own computer, and Amazon Polly has a free tier. ElevenLabs also has its own free plan with 10,000 credits a month.

What is the best open source alternative to ElevenLabs?

For plain English narration, Kokoro is a good place to start because it is small and Apache licensed. Piper is a fast engine with its own catalog of voices. Neither matches ElevenLabs for emotional range, and neither is a cloning tool.

Can I use the ElevenLabs free plan on monetized videos?

The ElevenLabs pricing page lists a commercial licence starting at the Starter plan. If you plan to earn from the videos, treat the free plan as a test and read ElevenLabs’ terms for the current rule.

Can a local voice model clone my voice?

Some can. Chatterbox clones from a reference clip, though it is no longer maintained. Kokoro’s model card lists fixed voices and does not describe a cloning feature. Only clone a voice you have the right to use.

Is ElevenLabs worth paying for?

If narration quality is what sets your channel apart, often yes. If you publish many short, clear, informational videos, a local voice may be good enough. In both cases the script matters more than the voice: see how to write a script for a faceless video.

Contact us

Licence keys, billing, bugs and ideas. We reply by email.