---
license: llama3.2
library_name: mlx
language:
  - en
  - fr
  - de
  - es
  - pt
  - it
  - pl
  - ja
  - zh
  - ar
tags:
  - mlx
  - tts
  - text-to-speech
  - speech-synthesis
  - tada
  - apple-silicon
pipeline_tag: text-to-speech
base_model: meta-llama/Llama-3.2-3B
arxiv: 2602.23068
---

<h1 align="center">TADA: A Generative Framework for Speech Modeling via Text-Acoustic Dual Alignment</h1>

<p align="center">
  <a href="https://arxiv.org/abs/2602.23068"><img src="https://img.shields.io/badge/arXiv-Paper-b31b1b.svg" alt="Paper"></a>
  <a href="https://huggingface.co/collections/HumeAI/tada"><img src="https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Collection-yellow" alt="Collection"></a>
  <a href="https://pypi.org/project/mlx-tada/"><img src="https://img.shields.io/badge/PyPI-mlx--tada-3775A9.svg?logo=pypi&logoColor=white" alt="PyPI"></a>
  <a href="https://www.hume.ai/blog/opensource-tada"><img src="https://img.shields.io/badge/Blog-Post-orange.svg" alt="Blog"></a>
</p>

<img width="2400" height="1260" alt="image" src="https://github.com/user-attachments/assets/800eb8c5-eb6f-4e03-b8f3-150055a6cdfc" />

<p align="center"><br/><em>A unified speech-language model that synchronizes speech and text into a single, cohesive stream via 1:1 alignment.</em></p>

---

# MLX-TADA-3B

Pre-converted [MLX](https://github.com/ml-explore/mlx) weights for [TADA](https://github.com/HumeAI/tada) (Text-Acoustic Dual Alignment) speech synthesis on Apple Silicon.

Built on [Llama 3.2 3B](https://huggingface.co/meta-llama/Llama-3.2-3B). Multilingual: English, French, German, Spanish, Portuguese, Italian, Polish, Japanese, Chinese, Arabic.

| Component | File | Size |
|-----------|------|------|
| LLM + VibeVoice head | `model/weights.safetensors` | 7.6 GB |
| Aligner | `aligner/weights.safetensors` | 852 MB |
| Decoder (DAC) | `decoder/weights.safetensors` | 226 MB |
| Encoder | `encoder/weights.safetensors` | 178 MB |
| **Total** | | **~8.9 GB** |

All weights are stored in bfloat16 safetensors format.

## Prerequisites

TADA models are built on [Meta Llama 3.2](https://huggingface.co/meta-llama). You must request access to the Llama models before using TADA:

- Visit [meta-llama/Llama-3.2-1B](https://huggingface.co/meta-llama/Llama-3.2-1B) or [meta-llama/Llama-3.2-3B](https://huggingface.co/meta-llama/Llama-3.2-3B) and accept the license agreement

## Quick Start

```bash
pip install mlx-tada
```

Or install from source:
```bash
git clone https://github.com/HumeAI/tada.git
cd tada/apple
uv venv && uv pip install -e .
```

Download a reference audio clip:
```bash
curl -O "https://storage.googleapis.com/hume_reference_speakers/ljspeech.wav"
```

### Python

```python
from mlx_tada import TadaForCausalLM, save_wav

model = TadaForCausalLM.from_pretrained("HumeAI/mlx-tada-3b", quantize=4)
ref = model.load_reference("ljspeech.wav")
out = model.generate("Hello, this is a test of TADA speech synthesis.", ref)
save_wav(out.audio, "output.wav")
```

## Offline Use

To download the weights locally for offline inference:

```python
from huggingface_hub import snapshot_download
snapshot_download("HumeAI/mlx-tada-3b", local_dir="./weights/3b")
```

Then load from the local path:
```python
model = TadaForCausalLM.from_weights("./weights/3b", quantize=4)
```

### CLI

```bash
python -m mlx_tada.generate \
  --weights ./weights/3b \
  --audio ljspeech.wav \
  --text "Hello, this is a test of TADA speech synthesis." \
  --quantize 4 \
  --output output.wav
```

## Hardware Requirements

| Precision | Memory |
|-----------|--------|
| bfloat16 (default) | ~17 GB |
| 4-bit quantized | ~6 GB |

4-bit quantization is recommended for most Apple Silicon devices — it is roughly 10x faster with 60% less memory and minimal quality loss. Machines with 16 GB unified memory should use quantized mode.


## Related

- [TADA GitHub](https://github.com/HumeAI/tada) — source code, PyTorch inference, training
- [TADA Paper](https://arxiv.org/abs/2602.23068) — arxiv
- [HumeAI/tada-3b-ml](https://huggingface.co/HumeAI/tada-3b-ml) — PyTorch weights
- [HumeAI/mlx-tada-1b](https://huggingface.co/HumeAI/mlx-tada-1b) — 1B English-only MLX weights
- [HumeAI/tada-codec](https://huggingface.co/HumeAI/tada-codec) — shared encoder, decoder, aligner weights


## 📚 Citation

If you use this project in your research, please cite our paper:

```bibtex
@article{dang2026tada,
  title={TADA: A Generative Framework for Speech Modeling via Text-Acoustic Dual Alignment},
  author={Dang, Trung and Rao, Sharath and Gupta, Ananya and Gagne, Christopher and Tzirakis, Panagiotis and Baird, Alice and Cłapa, Jakub Piotr and Chin, Peter and Cowen, Alan},
  journal={arXiv preprint arXiv:2602.23068},
  year={2026}
}
```

## Contact

Hume AI is an empathic AI research company. We research the datasets, tools, and models needed to give empathy to AI models to serve human wellbeing. If you're interested in any of our product or research collaborations, please reach out to us at hello@hume.ai

## Acknowledgements

This project is built using Llama 3.2.

Llama 3.2 is licensed under the Llama 3.2 Community License