inclusionAI / Ming-3.0-tts

Ming-3.0-tts
 Multilingual speech synthesis with free-form and paralinguistic control. 
 🤗 Hugging Face ｜ 🤖 ModelScope ｜ 📑 Blog & Demos ｜ 💬 DingTalk(钉钉) 
Ming-3.0-tts provides two speech-generation models: **0.5b**, a dense model, and **tiny**, a MoE model built on Ling V3 (7.9B total · 1.3B active).
## 🏆 Artificial Analysis Leaderboards
On the independent Artificial Analysis text-to-speech leaderboards (October 2026), Ming-3.0-tts ranks **1st among open-weights models on Controlled Voice**, **1st overall on Pronunciation Robustness**, and **3rd among open-weights models on Provider Voice**. Ranks change as new models are added.
| Leaderboard | Setting | Ming-3.0-tts | Rank |
| --- | --- | :---: | --- |
| Controlled Voice | Arena preference Elo; every model clones the same 8 voices (4 US, 4 UK) | 1069 | **1st among open-weights** · 13th overall |
| Pronunciation Robustness | Share of highlighted spans in English text that human reviewers judged correctly pronounced | 93.2% | **1st overall** |
| Provider Voice | Arena preference Elo; each model uses its own native voices | 1188 | **3rd among open-weights** · 17th overall |
Within Pronunciation Robustness, Ming-3.0-tts also ranks **1st** on Expanding Shorthand (text normalization, 96.5%) and Preserving Exact Sequences (sequence fidelity, 90.6%).
## Featured Abilities
- **Voice creation:** clone a reference voice, design one from text, or choose from 100+ Chinese and five English built-in voices.
- **Free-form control:** guide speech delivery with natural-language instructions.
- **Paralinguistic control:** insert 14 inline tags for laughter, sighs, and other vocal expressions in Chinese or English.
- **Languages and dialects:** synthesize Chinese, English, German, Spanish, French, Korean and Japanese, plus Sichuanese and Cantonese.
- **Emotion and attributes:** control emotion, or adjust speed, pitch and volume.
- **Multi-speaker dialogue:** generate a conversation from a script and speaker references (**tiny only**).
Capabilities use separate task modes, not arbitrary combinations. See the Cookbook for supported inputs, Performance for benchmarks, and Blog & Demos for audio examples.
## Demo Video
https://github.com/user-attachments/assets/4062ba4c-7d2d-4a45-a19b-760100b79f9c
## Architecture
Text and task-specific control tokens condition a language model with a DiT patch head and an audio tokenizer. Both models share this generation framework.
## News
- 🔥 **[2026.10]** Ming-3.0-tts-0.5b and Ming-3.0-tts-tiny are released, with inference code, a task cookbook and the evaluation tables below.
## Roadmap
- [x] Ming-3.0-tts-0.5b weights and inference code
- [x] Ming-3.0-tts-tiny weights and inference code
- [x] Ming-3.0-tts-tiny benchmarks
- [x] vLLM-Omni accelerated inference
- [ ] Technical report
## Contents
- Demo Video
- Models
- Quick Start
- Cookbook
- Built-in voices
- Performance
- Limitations
- Usage Disclaimer
## Models
 Model 
 Architecture 
 Parameters 
 Download 
 Ming-3.0-tts-0.5b 
 De