aiqbench analysis
DELTA-TTS addresses critical latency and hallucination issues in text-to-speech by adapting autoregressive models into discrete diffusion frameworks, enabling parallel token generation without sacrificing acoustic fidelity. This approach leverages lightweight LoRA fine-tuning to achieve significantly faster inference speeds while improving robustness against error propagation inherent in sequential decoding methods.
DELTA-TTS: Adapting Autoregressive Model into Diffusion Language Model for Text-to-Speech
arXiv:2607.04140v2 Announce Type: replace-cross Abstract: Autoregressive (AR) text-to-speech (TTS) models generate discrete speech tokens sequentially, which makes inference slow and can degrade robustness, since local errors propagate to later positions and can escalate into hallucination. This limitation stems from their left-to-right AR commitment: each token must be determined before future speech-token context is available. However, such ordering is not an inherent requirement for TTS, since the model receives the full input text before synthesis. In this paper, we introduce DELTA-TTS, a lightweight LoRA-based adaptation framework that converts a pretrained AR TTS model into a discrete diffusion language model (dLLM) for confidence-ordered speech-token decoding. To better capture the local structure of speech, DELTA-TTS incorporates a convolution module that injects local acoustic context, together with a 1/t-weighted training objective and a time-shifted inference schedule that together defer low-confidence positions to later steps. Trained on only 585 hours of LibriTTS, DELTA-TTS achieves a 1.75% WER on Seed-TTS test-en, outperforming its AR backbone while generating tokens 3.3x faster. Further analysis shows that DELTA-TTS produces sharper text--speech alignment, increases overall decoding confidence, and mitigates the hallucinations observed in AR generation.
Abstract:Autoregressive (AR) text-to-speech (TTS) models generate discrete speech tokens sequentially, which makes inference slow and can degrade robustness, since local errors propagate to later positions and can escalate into hallucination. This limitation stems from their left-to-right AR commitment: each token must be determined before future speech-token context is available. However, such ordering is not an inherent requirement for TTS, since the model receives the full input text before synthesis. In this paper, we introduce DELTA-TTS, a lightweight LoRA-based adaptation framework that converts a pretrained AR TTS model into a discrete diffusion language model (dLLM) for confidence-ordered speech-token decoding. To better capture the local structure of speech, DELTA-TTS incorporates a convolution module that injects local acoustic context, together with a 1/t-weighted training objective and a time-shifted inference schedule that together defer low-confidence positions to later steps. Trained on only 585 hours of LibriTTS, DELTA-TTS achieves a 1.75% WER on Seed-TTS test-en, outperforming its AR backbone while generating tokens 3.3x faster. Further analysis shows that DELTA-TTS produces sharper text--speech alignment, increases overall decoding confidence, and mitigates the hallucinations observed in AR generation.
Audio and Speech Processing (eess.AS); Computation and Language (cs.CL)
arXiv:2607.04140 [eess.AS]
(or arXiv:2607.04140v2 [eess.AS] for this version)
https://doi.org/10.48550/arXiv.2607.04140 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 278 words
Related coverage
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · Sep 25, 2026
Build a Physical AI model factory with NVIDIA Cosmos 3 on SageMaker HyperPod
AWS · Sep 4, 2026
UNIFUSION: Adapting Autoregressive Language Models into Discrete Diffusion under a Unified Reverse-Rate Objective
arXiv · Jul 28, 2026
FastGuide: Accelerating Reward Guidance for Diffusion Large Language Models
arXiv · Oct 1, 2026