{"ID":2868117,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.17021","arxiv_id":"2509.17021","title":"Bridging the gap between training and inference in LM-based TTS models","abstract":"Recent advancements in text-to-speech (TTS) have shown that language model (LM) based systems offer competitive performance compared to traditional approaches. However, in training, TTS models use ground-truth (GT) tokens as prefixes to predict the next token, while in inference these tokens are not available, a gap between training and inference that is often neglected. In this study, we propose a prompt-guided hybrid training scheme to mitigate exposure bias in popular LM-based TTS systems. Our core idea is to adopt a hybrid training paradigm that combines teacher forcing with free running, thereby introducing self-generated tokens into the training process. This makes the training mode more consistent with inference, reducing the training-inference gap. In addition, we incorporate an EOS prediction mechanism during training to detect incorrect sequence termination and adaptively control the free running process. Experimental results provide a comprehensive evaluation of the impact of exposure bias on LM-based TTS, and demonstrate that our method effectively narrows the training-inference gap, thereby improving the quality of synthesized long-form speech.","short_abstract":"Recent advancements in text-to-speech (TTS) have shown that language model (LM) based systems offer competitive performance compared to traditional approaches. However, in training, TTS models use ground-truth (GT) tokens as prefixes to predict the next token, while in inference these tokens are not available, a gap be...","url_abs":"https://arxiv.org/abs/2509.17021","url_pdf":"https://arxiv.org/pdf/2509.17021v1","authors":"[\"Ruonan Zhang\",\"Lingzhou Mu\",\"Xixin Wu\",\"Kai Zhang\"]","published":"2025-09-21T10:29:36Z","proceeding":"cs.SD","tasks":"[\"cs.SD\",\"eess.AS\"]","methods":"[\"Language Model\"]","has_code":false}