--- title: Ming-3.0-tts-tiny canonical_url: "https://www.modelscope.cn/models/inclusionAI/Ming-3.0-tts-tiny" md_url: "https://www.modelscope.cn/models/inclusionAI/Ming-3.0-tts-tiny.md" repository: inclusionAI/Ming-3.0-tts-tiny last_updated: 2026-10-19 license: mit pipeline_tag: text-to-speech tasks: - text-to-speech model_type: - lingv3_tts architectures: - LingV3TTSForConditionalGeneration library_name: - onnx - safetensors - pytorch frameworks: - pytorch language: - zh - en - de - es - fr - ko - ja downloads: 6 stars: 0 tags: - audio - text-to-speech - voice-cloning - multilingual - mixture-of-experts - speech-generation --- # Ming-3.0-tts-tiny > Ming-3.0-tts-tiny - inclusionAI 在 ModelScope 开源的模型。Ming-3.0-tts-tiny 7.9B-A1.3B multilingual MoE speech synthesis with paralinguistic and free-form control. inclusionAI/Ming-3.0-tts-tiny 是 ModelScope 魔搭社区上的text-to-speech模型，采用 mit 许可。 - **Repository**: inclusionAI/Ming-3.0-tts-tiny - **License**: mit - **Tasks**: text-to-speech - **Tags**: audio, text-to-speech, voice-cloning, multilingual, mixture-of-experts, speech-generation - **Downloads**: 6 - **Stars**: 0 - **Last updated**: 2026-10-19 Source: https://www.modelscope.cn/models/inclusionAI/Ming-3.0-tts-tiny ---
Ming-3.0-tts Logo

Ming-3.0-tts-tiny

7.9B-A1.3B multilingual MoE speech synthesis with paralinguistic and free-form control.

GitHub | Blog & Demos | ModelScope

This repository contains **Ming-3.0-tts-tiny**, the Ming-3.0-tts mixture-of-experts checkpoint built on Ling V3. Its Ling V3 backbone has 7.9B total parameters and activates about 1.3B parameters per token. The repository includes sharded weights, supporting model code, audio tokenizer, speaker encoder and text tokenizer. > **Inference requires the [Ming-3.0-tts GitHub repository](https://github.com/inclusionAI/Ming-3.0-tts).** This Hugging Face repository is for checkpoint distribution and does not provide a standalone inference setup. Clone the GitHub repository and follow its README for the current dependencies, inference code and usage instructions. The CLI scripts and Python APIs referenced below are provided by the GitHub repository. ## Model Details | Item | Value | | --- | --- | | Architecture | Mixture of Experts, Ling V3 backbone | | Backbone parameters | 7.9B total, about 1.3B active per token | | Languages | Chinese, English, German, Spanish, French, Korean, Japanese | | Dialects | Sichuanese, Cantonese | | Python API | `ming_tiny.MingTinyTTS` | | CLI | `infer_tiny.py` | | Prompt protocol | `thinking_v1` | ## Capabilities - Zero-shot voice cloning from a short reference clip - Voice Design from a natural-language speaker description - Free-form control of delivery and speaking style - 100+ built-in Chinese voices and five built-in English voices - Multilingual synthesis and Sichuanese/Cantonese dialect control - 14 inline tags for breath, cough, sigh, laughter, hesitation, surprise and question/confirmation sounds in Chinese or English - Explicit emotion, speed, pitch and volume control - Multi-speaker dialogue from a script and speaker references (`podcast`)

Ming-3.0-tts-tiny architecture

## Quick Start ### Installation ```bash git clone https://github.com/inclusionAI/Ming-3.0-tts.git cd Ming-3.0-tts conda create -n ming-tiny python=3.10.15 -y conda activate ming-tiny pip install -r requirements-tiny.txt ``` Tiny uses a separate reference inference environment: Python **3.10.15**, PyTorch and torchaudio **2.7.1+cu118** (CUDA 11.8 wheels), and `transformers==4.46.3`. Install the pinned dependencies from [requirements-tiny.txt](https://github.com/inclusionAI/Ming-3.0-tts/blob/main/requirements-tiny.txt) in a dedicated environment, as other dependency versions differ from 0.5B. We recommend using NVIDIA H800 for inference. ### Download ```bash pip install "huggingface_hub[cli]" hf download inclusionAI/Ming-3.0-tts-tiny --local-dir Ming-3.0-tts-tiny ``` ### CLI Inference ```bash python infer_tiny.py \ --model_path inclusionAI/Ming-3.0-tts-tiny \ --task dialect \ --text "今日天气几好，我哋食完饭一齐去海边行下啦。" \ --dialect 广粤话 \ --prompt_wav data/wavs/10002287-00000094.wav \ --output_path output/tiny.wav ``` ### Python API Loading configures the release runtime (bf16, portable MoE loop, fp32 recurrent state) and checks the checkpoint against it; `generate()` defaults to `cfg_scale=2.0`, `temperature=0`, `flow_steps=10`, and `sigma=0.25`. ```python from ming_tiny import MingTinyTTS tts = MingTinyTTS.from_pretrained("inclusionAI/Ming-3.0-tts-tiny", device="cuda:0") tts.generate( task="seed", text="傍晚的风从山口吹过来，远处的云慢慢染成了金色。", prompt_wav="data/wavs/10002287-00000094.wav", prompt_text="在此奉劝大家别乱打美白针。", target_language="Chinese", # English for English text output_path="outputs/seed.wav", ) ``` ### Supported Tasks Each call selects one task and accepts the inputs listed below: | Task | Reference audio | Reference text | `target_language` | Controls | | --- | --- | --- | --- | --- | | `seed` | one | required | `Chinese` / `English`, required | — | | `multilingual` | one | required | `German` / `Spanish` / `French` / `Korean` / `Japanese` | — | | `paralinguistic` | one | required | `Chinese` / `English` | inline tags in `text` | | `podcast` | one per speaker | one per speaker, same order | inferred from the text | `speaker_N:` at the start of every line | | `basic` | one | not accepted | — | `speed`, `pitch`, `volume`, all three | | `emotion` | one | not accepted | — | `emotion` | | `dialect` | one | not accepted | — | `dialect`: `川渝话` / `广粤话` | | `freeform` | one | not accepted | — | `instruction` | | `voice_design` | none | — | — | `style` | | `ip` | none | — | — | `ip` | Freeform always requires a reference clip and cannot be combined with voice design, a built-in voice, emotion, dialect or attribute controls. For `podcast`, start every dialogue line with `speaker_N:` and provide one reference audio clip and transcript per speaker, in speaker order. Repeat `--prompt_wav` and `--prompt_text` on the command line. See the [Tiny examples](https://github.com/inclusionAI/Ming-3.0-tts#ming-30-tts-tiny) for a Podcast Python example and [examples/one_click/](https://github.com/inclusionAI/Ming-3.0-tts/tree/main/examples/one_click) for scripts for every task. `python infer_tiny.py --help` lists the CLI arguments. ### Verify the Checkpoint ```bash python tools/preflight_tiny.py --model_path /path/to/Ming-3.0-tts-tiny ``` ## Performance The tables below contain results for **this Tiny checkpoint only**. See the [project README](https://github.com/inclusionAI/Ming-3.0-tts#performance) for evaluation details and full cross-model comparisons. ### Zero-shot TTS (Seed-TTS testset) | zh WER ↓ | zh SIM ↑ | en WER ↓ | en SIM ↑ | | ---: | ---: | ---: | ---: | | 0.820% | 0.781 | 1.666% | 0.728 | ### Multilingual (MiniMax test set, selected languages) Evaluated on the German, Spanish, French, Korean, and Japanese subsets of [MiniMaxAI/TTS-Multilingual-Test-Set](https://huggingface.co/datasets/MiniMaxAI/TTS-Multilingual-Test-Set). | Language | WER/CER ↓ | SIM ↑ | | --- | ---: | ---: | | German | 0.325% | 0.791 | | Spanish | 0.971% | 0.797 | | French | 3.470% | 0.733 | | Korean | 6.000% | 0.797 | | Japanese | 3.427% | 0.780 |

Each open-source model was evaluated using its recommended inference settings under a unified ASR-based WER/CER evaluation protocol.

### Paralinguistic Control (NVBench) | Split | Language | CER ↓ | PCER ↓ | OCER ↓ | | --- | --- | ---: | ---: | ---: | | Single | Chinese | 2.80 | 17.23 | 3.20 | | Single | English | 5.86 | 32.95 | 7.05 | | Multi | Chinese | 4.49 | 32.88 | 5.81 | | Multi | English | 5.12 | 46.13 | 7.68 | ### Dialect Control | Test set | CER ↓ | SIM ↑ | ACC ↑ | | --- | ---: | ---: | ---: | | WSC-Eval-TTS-easy | 2.53% | 0.721 | 86.09% | | WSC-Eval-TTS-hard | 3.11% | 0.738 | 88.94% | | WSYue-TTS-eval-Base | 7.40% | 0.605 | 92.90% | | WSYue-TTS-eval-Coverage | 8.67% | 0.653 | 95.16% | ### Emotion, Attribute Control and Voice Design | Evaluation | Result | | --- | --- | | CV3-Eval emotion average | 74.00% | | CV3N hard emotion average | 30.67% | | Attribute control: speed / volume / pitch / average | 93.00% / 93.00% / 78.70% / 88.23% | | Attribute control: WER / SIM | 0.274% / 0.742 | | Voice Design: APS / DSD / RP / average | 85.10% / 79.90% / 61.10% / 75.37% | ### Podcast | CER ↓ | cpSIM ↑ | UTMOS ↑ | | ---: | ---: | ---: | | 2.03% | 0.561 | 2.12 | ## Limitations - Korean CER (6.000%) is the clearest multilingual content-accuracy gap for this checkpoint. - Speaker similarity remains weaker than content accuracy on several multilingual and dialect evaluations. - NVBench Multi trails Single on several metrics; stability with multiple paralinguistic events in one sentence is still improving. - Emotion control is harder when the words carry no emotional signal. - Pitch is less accurate than speed and volume, and open-ended Voice Design does not yet match the strongest evaluated system. - Podcast audio quality remains a limitation: UTMOS is 2.12, below ZipVoice-Dia's 2.24 and the historical Ming-omni-tts-0.5B score of 2.25. ## Usage Disclaimer - Voice cloning and built-in character voices are provided for research and legitimate creative use. Do not use them to impersonate a real person without consent, to deceive, or for any illegal purpose. - Label synthetic speech as synthetic where it could be mistaken for a real recording. - The authors accept no liability for misuse. If you find abuse of this project, please report it to us. ## Acknowledgements - Ling V3 for the mixture-of-experts backbone - [CAM++](https://modelscope.cn/models/iic/speech_campplus_sv_zh-cn_16k-common) for speaker embedding extraction - [Ming-omni-tts](https://github.com/inclusionAI/Ming-omni-tts), the previous generation, whose audio tokenizer this release builds on ## License Released under the [MIT license](https://github.com/inclusionAI/Ming-3.0-tts/blob/main/LICENSE).