FREEDOM251

Freedom251 › Text-to-Speech Software

MARS5-TTS

Free plan APILinuxSelf-hosted

Visit MARS5-TTS Alternatives Pricing

MARS5-TTS's own site

What MARS5-TTS is

MARS5-TTS is CAMB.AI’s free, open-source English text-to-speech model. It generates speech from text with a short audio reference for voice cloning, including prosodically difficult material such as sports commentary and anime. Punctuation and capitalization can guide pauses and emphasis. Shallow cloning works without a reference transcript; deep cloning uses one and typically gives higher quality with slower inference. The project recommends clean reference audio from one to 12 seconds, with about six seconds as an optimal length. Its two-stage pipeline combines autoregressive and non-autoregressive processing with multinomial diffusion to refine speech codes. MARS5-TTS supports API, Linux, and self-hosted use, and commercial use is marked yes. Local setup requires Python 3.10 or later, Torch 2.0 or later, and packages including Torchaudio, Librosa, Vocos, Encodec, safetensors, and regex. GPU inference requires capacity for at least 750M plus 450M parameters and 750M active parameters.

Who it suits

MARS5-TTS suits developers building English speech generation or voice-cloning workflows through an API, Linux machine, or self-hosted deployment. It also fits projects able to meet its software and GPU requirements.

What it does well

  • Free open-source English model
  • Shallow and deep cloning modes
  • API and self-hosted deployment options
  • Commercial use marked yes
  • Docker and Hugging Face documentation

What to know first

  • Requires substantial GPU capacity locally
  • Deep cloning is slower
  • Project lists inference stability improvements needed
  • Uses GNU AGPL 3.0

Plans

PlanBilled and limitsPrice
MARS5-TTSEnglish speech model; AGPL-3.0; commercial licensing inquiries by emailFree

As github.com lists them, read 12 Oct 2026. The maker's own page has the final word.

Verdict

MARS5-TTS offers reference-based English speech generation with two cloning modes and multiple deployment paths. Its hardware demands, AGPL license, and listed development limitations matter for local or commercial planning.

Facts from MARS5-TTS's own pages

API access
The README directs users without the required hardware to use the CAMB.AI API and says they can contact [email protected] about extra test credits. · github.com, 12 Oct 2026
Commercial use
Yes · github.com, 20 Sept 2026
Deployment
The project documents loading through torch.hub or Hugging Face and provides a Docker image and Dockerfile build instructions. · github.com, 12 Oct 2026
Hardware limit
Local inference requires GPU capacity to store at least 750M plus 450M parameters and run inference with 750M active parameters. · github.com, 12 Oct 2026
Inference modes
MARS5 supports shallow cloning without a reference transcript and deep cloning with one, which typically provides higher quality at the cost of slower inference. · github.com, 12 Oct 2026
Known limitations
The project lists inference stability, speed, long-reference selection, and long-form generation among areas needing improvement or contribution. · github.com, 12 Oct 2026
License
The English model is open-sourced under GNU AGPL 3.0, and the README says to email CAMB.AI about commercial inquiries or licensing the closed-source version. · github.com, 12 Oct 2026
Local requirements
The quickstart lists Python 3.10 or later and Torch 2.0 or later, plus Torchaudio, Librosa, Vocos, Encodec, safetensors, and regex. · github.com, 12 Oct 2026
Platforms
api,self_hosted · github.com, 20 Sept 2026
Prosody controls
Punctuation and capitalization in the text can guide pauses and emphasis in generated speech. · github.com, 12 Oct 2026
Purpose
MARS5 is CAMB.AI’s English text-to-speech speech model. · github.com, 12 Oct 2026
Reference audio
The README recommends clean reference audio between 1 and 12 seconds and says around 6 seconds gives optimal results. · github.com, 12 Oct 2026
Speech generation
The model can generate speech from text using a short audio reference, including prosodically challenging scenarios such as sports commentary and anime. · github.com, 12 Oct 2026
Support
The repository invites questions and feedback through the CAMB.AI forum and Discord, and lists [email protected] for API credit inquiries. · github.com, 12 Oct 2026
Technical design
MARS5 uses a two-stage autoregressive and non-autoregressive pipeline, with a multinomial diffusion component refining speech codes. · github.com, 12 Oct 2026
Voice cloning
Speaker identity is specified with an audio reference, and supplying its transcript enables a slower deep-clone mode intended to improve cloning quality. · github.com, 12 Oct 2026

Where MARS5-TTS stands on Freedom251

Look at these too