Search papers, labs, and topics across Lattice.
The dots.tts model is a 2B-parameter continuous autoregressive text-to-speech (TTS) foundation model that innovatively employs a semantically structured continuous speech space through a trained AudioVAE, full-history conditioning for long-range consistency, and reward-free self-corrective post-training. This model achieves state-of-the-art performance on the Seed-TTS-Eval benchmark, with remarkably low word error rates (WERs) and high similarity scores across multiple languages. Additionally, it enables efficient inference with low-latency speech generation, making it a robust choice for practical applications in TTS systems.
Achieving the best average performance on Seed-TTS-Eval, dots.tts sets a new standard for multilingual text-to-speech systems with unprecedented robustness and emotional expressiveness.
We present dots.tts, a 2B-parameter continuous autoregressive text-to-speech (TTS) foundation model that models speech in a continuous latent space. Compared with existing continuous autoregressive models, our key innovations are threefold. First, we train an AudioVAE with multiple objectives to build a semantically structured and prediction-friendly continuous speech space. Second, we use full-history conditioning in the flow-matching head to preserve long-range consistency and reduce drift during generation. Third, we apply reward-free self-corrective post-training to the flow-matching head to further improve robustness and acoustic quality. After being trained on a large-scale multilingual corpus, dots.tts achieves the best average performance on Seed-TTS-Eval, with WERs of 0.94%/1.30%/6.60% and SIM scores of 81.0/77.1/79.5 on the zh/en/zh-hard test sets, respectively. Across other benchmarks, dots.tts also consistently demonstrates open-source state-of-the-art performance, exhibiting strong generation stability, voice cloning ability, and emotional expressiveness. For efficient inference, we further apply CFG-aware MeanFlow distillation, enabling low-latency speech generation with first-packet latencies of 85/54 ms in output streaming and dual-streaming modes, respectively. To facilitate reproducible research and practical deployment, we release the training and inference code, together with the pretrained, post-trained, and MeanFlow-distilled checkpoints, under the Apache 2.0 license.
