The paper introduces a method for improving fine-grained, temporally aligned outputs in Speech Large Language Models (SpeechLLMs) by replacing absolute timestamps with relative timestamps, which reduces vocabulary size and enhances generalization. It proposes a hybrid fine‑tuning strategy that fully fine‑tunes the timestamp‑augmented embedding layer and language model head while applying LoRA to decoder layers, and introduces a masked timestamp training objective to prevent over‑reliance on ground‑truth timestamps. Experiments show significant gains in timestamp prediction accuracy without compromising transcription quality.
By Quanwei Tang, Zhiyu Tang, Xu Li, Dong Zhang, Shoushan, Guodong Zhou
arXiv:2607. 11889v1 Announce Type: cross Abstract: Large language models trained on unrestricted internet corpora inevitably embed information from the future, introducing lookahead bias that compromises the validity of backtests and causal inference in finance and the social sciences.
By Bryan Kelly, Semyon Malamud, Johannes Schwab, Teng Andrea Xu
arXiv:2602. 10230v2 Announce Type: replace Abstract: Audio language models process input audio into rich frame-level representations, but the standard approach to temporal localization generates timestamps as sequences of text tokens, which discards the frame-level representations in favor of autoregressive decoding.
By Joseph An, Phillip Keung, Jiaqi Wang, Orevaoghene Ahia, Noah A. Smith
arXiv:2606. 01802v1 Announce Type: cross Abstract: MOSS-Audio is a unified audio-language model for speech, environmental sound, and music understanding, supporting audio captioning, time-aware question answering, timestamped transcription, and audio-grounded reasoning.
By Chen Yang, Chufan Yu, Hanfu Chen, Jie Zhu, Jingqi Chen, Ke Chen, Wenxuan Wang, Yang Wang, Yaozhou Jiang, Yi Jiang, Zhengyuan Lin, Ziqi Chen, Zhaoye Fei, Chenghao Liu, Jun Zhan, Kang Yu, Kexin Huang, Mingshu Chen, Qinyuan Cheng, Ruixiao Li, Shimin Li, Songlin Wang, Yang Gao, Yiyang Zhang, Xipeng Qiu
arXiv:2607. 05364v1 Announce Type: cross Abstract: Modern autoregressive ASR systems can emit timestamps as decoded tokens, enabling timestamped transcription without frame-level aligners or inference-time post-processing.
By Cheng-Kang Chou, Ming-To Chuang, Ke-Han Lu, Chan-Jan Hsu, Hung-yi Lee
Modern autoregressive ASR systems can emit timestamps as decoded tokens, enabling timestamped transcription without frame-level aligners or inference-time post-processing. We show that these generated timestamps can drift across long non-speech spans: the transcript may remain plausible, but the decoded time axis drifts away from the audio.
arXiv:2601. 14968v2 Announce Type: replace-cross Abstract: Most existing time series classification methods adopt a discriminative paradigm that maps input sequences directly to one-hot encoded class labels.
By Mingyue Cheng, Xiaoyu Tao, Huajian Zhang, Qi Liu, Zhiding Liu, Yucong Luo, Yiheng Chen, Enhong Chen
Modern ASR models trained on heterogeneously annotated data treat transcription style (verbatim vs. intended) as an uncontrolled latent variable, causing measurable decoding instability, evaluation confounding (up to 60% of reported WER attributable to style mismatch), and unreliable word-level timing.
arXiv:2603. 05299v2 Announce Type: replace-cross Abstract: Large language models show that simple autoregressive training can yield scalable and coherent generation, but extending this paradigm to speech remains challenging due to the entanglement of semantic and acoustic information.
By Luca Della Libera, Cem Subakan, Mirco Ravanelli
arXiv:2608. 08569v1 Announce Type: new Abstract: Recent advancements in Speech Large Language Models have demonstrated remarkable capabilities in understanding complex audio tasks.
By Wenxu Jia, Dongjie Fu, Xize Cheng, Fangming Feng, Linjun Li, Wenshi Chen, Yingming Li, Zhou Zhao, Tao Jin
arXiv:2606. 02991v1 Announce Type: cross Abstract: We introduce TypewriterLM, a 7.
By Xiaoxi Luo, Zachary Shinnick, Niclas Griesshaber, Yixuan Wang, Junchi Yu, Freda Shi, Philip Torr, Yao Lu
Conversational ASR for lower-resource languages and niche domains is limited by the scarcity of domain-matched multi-speaker training data. We propose an augmentation pipeline that generates scenario-level dialogues with participant metadata, maps speaker attributes to TTS voice profiles, and assembles synthesized utterances into speaker-aware simulated conversations.