arXiv subjects

Nikolaos Lagos

Publications and source records attributed to Nikolaos Lagos.

3 recordsLinked to original sources

arXiv2026

StarDrinks: An English and Korean Test Set for SLU Evaluation in a Drink Ordering Scenario

LLMs and speech assistants are increasingly used for task-oriented interactions, yet their evaluation often relies on controlled scenarios that fail to capture the variability and complexity of real user requests. Drink ordering, for example, involves diverse named entities, drink types, sizes, customizations, and brand-specific terminology, as well as spontaneous speech phenomena such as hesitations and self-corrections. To address this gap, we introduce StarDrinks, a test set in English and Korean containing speech utterances features, transcriptions, and annotated slots. Our dataset supports speech-to-slots SLU, transcription-to-slots NLU, and speech-to-transcription ASR evaluation, providing a realistic benchmark for model robustness and generalization in a linguistically rich, real-world task.

cs.CL↗

arXiv2025

Impurities in cryogenic solids: a new platform for hybrid quantum systems

Hybrid quantum systems offer a promising platform for studying quantum phenomena and developing applied technologies, benefiting from the individual strengths of their components. Here, we present a novel hybrid quantum platform composed of solid noble gas crystals doped with spin impurities atop superconducting resonators. The noble gas crystals provide a soft, inert, predominantly spin-0 host matrix for the atomic impurities, while the alkali atoms have addressable and long-lived hyperfine transitions in the GHz regime. We demonstrate the ability to reach the strong coupling regime between the atomic impurity ensemble and the superconducting resonator at mK temperatures, and perform coherence time measurements. Our proof-of-principle measurements show that this platform offers a unique architecture for exploring fundamental quantum effects and new quantum technologies.

quant-ph↗

arXiv2024

mHuBERT-147: A Compact Multilingual HuBERT Model

We present mHuBERT-147, the first general-purpose massively multilingual HuBERT speech representation model trained on 90K hours of clean, open-license data. To scale up the multi-iteration HuBERT approach, we use faiss-based clustering, achieving 5.2x faster label assignment than the original method. We also apply a new multilingual batching up-sampling strategy, leveraging both language and dataset diversity. After 3 training iterations, our compact 95M parameter mHuBERT-147 outperforms larger models trained on substantially more data. We rank second and first on the ML-SUPERB 10min and 1h leaderboards, with SOTA scores for 3 tasks. Across ASR/LID tasks, our model consistently surpasses XLS-R (300M params; 436K hours) and demonstrates strong competitiveness against the much larger MMS (1B params; 491K hours). Our findings indicate that mHuBERT-147 is a promising model for multilingual speech tasks, offering an unprecedented balance between high performance and parameter efficiency.

cs.CL↗