Breeze TTS 2 is the #1 Open-Weight TTS on Artificial Analysis Arena Leaderboard.

PlaygroundRead the blog
Breeze Blue

The most engaging voice AI for real-time interaction

Create distinctive voices, direct every nuance, and deliver expressive speech in 50+ languages with industry-leading low latency. Built for real-time experiences, content production, and voice applications at scale.

Unlock your full audio experienceOpen App

Independent Evaluation

#1 Open-Weight TTS

Breeze TTS 2 ranks first among open-weight models and above leading proprietary models on Artificial Analysis' Provider Voice Arena, based on blind listener preferences.

Text-to-Speech Models by Elo Score

BreezeBlueOpen weightsClosed weights
  1. #1
    BreezeBlue · Breeze TTS 21215
  2. #2
    Google · Gemini 3.1 Flash TTS1208
  3. #3
    StepFun · StepAudio 2.5 TTS1205
  4. #4
    Cartesia · Sonic 3.51193
  5. #5
    ElevenLabs · Eleven v31175
  6. #6
    MiniMax · Speech 2.8 HD1171
  7. #7
    Gradium · Gradium TTS1149
  8. #8
    Fish Audio · Fish Audio S2 Pro1128
  9. #9
    Mistral · Voxtral TTS1078
  10. #10
    Boson AI · Higgs Audio V3 TTS1038
  11. #11
    Resemble AI · Chatterbox1021

Source: Artificial Analysis Text-to-Speech Provider Voice Arena Leaderboard (2026-09-08)

Model

Create the voice you imagined

Design a voice, shape its performance, and generate speech in real time across multiple languages

From a prompt to the right voice

Create voices that match the role and span a wider range of characters.

Try Voice Design
View Voice Design Benchmark
Voice Design — descriptive tags like gender, age, tone, and language flowing into a generated voice card

BreezeCreator

Turn your ideas into living voices, within one powerful AI platform

Generate directed speech, or choose from a library of voices with real character. Produce long-form content like podcasts and audiobooks — and clone or design your own voice whenever a project calls for it.

Start Creating
Text to Speech demo — say how a line should sound and the voice steers to match

Steer every line with words

Say how it should sound, and the voice steers to match.

Find the right voice for your scene

A diverse library of voices, each with a personality of its own.

Explore Voice Library
Make your own voice — clone a voice from audio or design a new one from a prompt

Make your own voice

Clone a voice from audio, or design a new one from a prompt.

BreezeAPI

Build voice products with Breeze API

Build with the Breeze Developer API. Use the same Breeze account, subscription plan, credit balance, and saved voices across the API and Studio. Manage API keys, API usage, and request logs in the Developer Console.

Read DocsGet API Key

Clone and use a voice

Clone a voice from a reference audio sample and use it across your projects. Maintain consistency while adapting it to different scenarios.

Design a voice

Design a voice from scratch using a text prompt. Define the character, style, and qualities without recording any audio.

Direct a performance

Direct a performance by describing how the voice should sound. Control tone, emotion, and delivery to shape each line.

Latest Updates

Stay up to date with our latest work, research, and collaborations.