A unified speech-language model that synchronizes speech and text into a single, cohesive stream via 1:1 alignment.
Text-Acoustic Dual-Alignment Large Language Model
TADA is a unified speech-language model that synchronizes speech and text into a single, cohesive stream via 1:1 alignment. By leveraging a novel tokenizer and architectural design, TADA achieves high-fidelity synthesis and generation with a fraction of the computational overhead required by traditional models.
1:1 Token Alignment: Unlike standard models, TADA’s tokenizer encodes audio into a sequence of vectors that perfectly matches the number of text tokens.
Dynamic Duration Synthesis: As a TTS model, it generates the full speech segment for a text token in a single autoregressive step, regardless of length. This eliminates the need for fixed-frame-rate processing.
Dual-Stream Generation: In speech-language modeling mode, it generates a text token and the speech for the preceding token simultaneously, maintaining the same context length and minimal overhead compared to text-only generation.
Efficiency & Reliability: TADA delivers superior expressiveness and natural flow while significantly reducing the computational cost associated with fixed audio frame rates.
How It Works
The Tokenization Schema
TADA unifies modalities by ensuring that for every word or subword token, there is exactly one corresponding speech vector. This synchronized stream allows the model to "understand" the precise timing of speech relative to text.
Dynamic Autoregression
Most TTS models require a fixed number of steps to produce one second of audio (e.g., 50 frames per second). TADA breaks this constraint:
Each autoregressive step covers one text token.
The model dynamically determines the duration and prosody for that specific token.
This results in a more natural flow and eliminates transcript hallucination.
All models use the same encoder (HumeAI/tada-codec) and can be loaded using the same API.
Evaluation
CER
Speed
Speaker Similarity
MOS
Run Inferece
Text-to-speech
python
1import torch
2import torchaudio
34from tada.modules.encoder import Encoder
5from tada.modules.tada import TadaForCausalLM
67device ="cuda"8encoder = Encoder.from_pretrained("HumeAI/tada-codec", subfolder="encoder").to(device)9model = TadaForCausalLM.from_pretrained("HumeAI/tada-1b").to(device)1011audio, sample_rate = torchaudio.load("samples/ljspeech.wav")12audio = audio.to(device)13prompt_text ="The examination and testimony of the experts, enabled the commission to conclude that five shots may have been fired."14prompt = encoder(15 audio, text=[prompt_text], sample_rate=sample_rate
16)1718output = model.generate(19 prompt=prompt,20 text="Please call Stella. Ask her to bring these things with her from the store.",21)
Speech continuation
Provide num_extra_steps if you want to generate text+speech continuation of the prompt
If you use this project in your research, please cite our paper:
bibtex
1@article{dang2026tada,
2 title={TADA: A Generative Framework for Speech Modeling via Text-Acoustic Dual Alignment},
3 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},
4 journal={arXiv preprint arXiv:2602.23068},
5 year={2026}
6}
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.
