# Kyutai TTS (tts-1.6b-en\_fr)

> Open-weights streaming text-to-speech model with ultra-low latency for English and French

Kyutai TTS is an open-weights, 1.8B-parameter streaming speech model designed for real-time voice applications. It begins generating audio after just a few input words and runs with minimal latency, supporting high-throughput batching on single GPUs. The weights are released under a permissive CC-BY-4.0 license, making it suitable for production conversational agents.

Website: https://huggingface.co/kyutai/tts-1.6b-en_fr

## What it is for

- Synthesize real-time voice responses for conversational AI agents
- Stream live text-to-speech outputs in French and English
- Deploy multi-user low-latency voice services on self-hosted GPU hardware
- Apply custom pre-computed voice embeddings for varied speaker styles

## Features

- Hierarchical Transformer architecture (1B backbone, 600M depth transformer) using Mimi audio tokens
- Low-latency streaming generation that starts producing audio before reading the entire prompt
- Trained on 2.5 million hours of audio with CFG distillation for faster inference
- Supports concurrent batching up to 75x realtime generation throughput
- Voice conditioning via pre-computed cross-attention embeddings
- Bilingual support for English and French speech

## Pricing

Open source. Free, open-source under CC-BY-4.0

## Details

- Type: AI model
- Category: [AI](https://xiry.app/topics/ai.md)
- Topic: [Text to Speech](https://xiry.app/topics/ai/text-to-speech.md)
- Open source: yes
- Platforms: Linux, Self-hosted, Code library
- For: Developers and AI researchers building real-time conversational voice products
- Alternative to: ElevenLabs, OpenAI TTS, Cartesia, Kokoro TTS
- Tags: tts, streaming audio, voice ai, moshi, speech synthesis, open weights

## Why it was saved

- [@reach_vb, 3 July 2025](https://x.com/reach_vb/status/1940786922546249961): Open-license, low-latency streaming TTS for building real-time voice products.

## Compare

- [Alternatives to ElevenLabs](https://xiry.app/alternatives/elevenlabs.md)

## Related

- [ElevenLabs](https://xiry.app/resources/elevenlabs): AI voice generator, voice cloning, and conversational agent platform (Web app, Freemium)
- [Cartesia Sonic](https://xiry.app/resources/cartesia-sonic): Ultra-low-latency streaming text-to-speech and instant voice cloning model. (AI model, Freemium)
- [Dia](https://xiry.app/resources/dia): Open-source 1.6B text-to-speech model generating multi-speaker dialogue in one pass. (AI model, Open source)
- [Airy](https://xiry.app/resources/airy): Fast, low-cost text-to-speech API and studio for creators and developers (AI model, Freemium)
- [AudioPen](https://xiry.app/resources/audiopen): Converts spoken thoughts into structured, polished writing in customizable styles. (Web app, Freemium)
- [cartesia-python](https://xiry.app/resources/cartesia-python): Official Python library for Cartesia's real-time text-to-speech API (GitHub repo, Paid)

Updated 10 October 2026. Page: https://xiry.app/resources/kyutai-tts-tts-1-6b-en-fr

Make your own library: Xiry takes your saved reels, tweets, Reddit bookmarks and Watch Later videos, makes them searchable and pulls the useful stuff out, like this library. It's invite-only for now; join the waitlist at https://xiry.app/#waitlist.