%Aigaion2 BibTeX export from Idiap Publications
%Saturday 17 October 2026 03:54:19 PM

@INPROCEEDINGS{Kumar_SLT2026-2_2026,
                      author = {Kumar, Shashi and Villatoro-Tello, Esa{\'{u}} and Burdisso, Sergio and Hacioğlu, Kadri and Ba{\~{n}}eras-Roux, Thibault and Watawana, Hasindri Sankalpana and Sanchez-Cortes, Dairazalia and Madikeri, Srikanth and Motlicek, Petr and Stolcke, Andreas},
                    projects = {UNIPHORE, ELOQUENCE},
         mainresearchprogram = {AI for Everyone},
  additionalresearchprograms = {AI for Life},
                       title = {Abstract Compression of Conversational Audio Context for LLM-based ASR},
                   booktitle = {IEEE Spoken Language Technology 2026},
                        year = {2026},
                         url = {https://arxiv.org/abs/2603.26246},
                    abstract = {Standard LLM-based speech recognition systems process utterances in isolation, limiting their ability to leverage conversational context. Here we study whether multimodal context from prior turns improves LLM-based ASR and how to represent that context efficiently. We find that, after supervised multi-turn training, conversational context mainly helps with the recognition of contextual entities. However, conditioning on raw context is expensive because the prior-turn audio embedding sequence grows rapidly with conversation length. To address this, we propose Abstract Compression, a learned latent abstraction of prior-turn audio that replaces each prior-turn audio sequence with a fixed number of latent tokens while retaining corresponding transcripts explicitly. On both in-domain and out-of-domain test sets, the compression-enabled model recovers part of the gains of raw-context conditioning with a smaller prior-turn audio footprint. We also provide targeted analyses of the compression setup and its tradeoffs.},
                         pdf = {https://publications.idiap.ch/attachments/papers/2026/Kumar_SLT2026-2_2026.pdf}
}