Free-form conversation with simultaneous speaker recognition on mobile robots

Determine whether free-form conversation can be handled appropriately by a mobile robot equipped with simultaneous speaker recognition in multi-party interaction settings.

Background

The paper discusses prior work integrating simultaneous speaker recognition into mobile robots. It notes that this research predates LLMs and therefore leaves unresolved whether such systems can support unrestricted, free-form conversation rather than only earlier dialogue paradigms or constrained interaction tasks.

This problem is relevant to the paper’s objective of enabling multi-party human–robot conversation in noisy public environments, where mobile robots must recognize multiple speakers and sustain conversation beyond fixed or narrowly defined exchanges.

References

Conversely, simultaneous speaker recognition has been integrated into a mobile robot , but this was before the development of LLMs and it is unknown if free-form conversation could be handled appropriately.

— Human-robot conversation with multiple participants in noisy public spaces  (2609.00648 - Lala et al., 1 Sep 2026) in Section II, Related Work

The first edition showed that current Speech LLMs largely solve transcription accuracy, while speaker diarization and deeper conversational understanding remained open; building on these findings, the second edition broadens language coverage and shifts emphasis toward diarization, acoustic understanding, and semantic understanding of full conversations.

— The Eloquence submission for Task 2 of the Interspeech 2026 MLC-SLM challenge  (2609.11724 - Luque et al., 10 Sep 2026) in Introduction