Kunnafonidilaw ka Cadeau: an ASR dataset of present-day Bambara

We present Kunkado, a 160-hour Bambara ASR dataset compiled from Malian radio archives to capture present-day spontaneous speech across a wide range of topics. It includes code-switching, disfluencies, background noise, and overlapping speakers that practical ASR systems encounter in real-world use. We finetuned Parakeet-based models on a 33.47-hour human-reviewed subset and apply pragmatic transcript normalization to reduce variability in number formatting, tags, and code-switching annotations. Evaluated on two real-world test sets, finetuning with Kunkado reduces WER from 44.47\% to 37.12\% on one and from 36.07\% to 32.33\% on the other. In human evaluation, the resulting model also outperforms a comparable system with the same architecture trained on 98 hours of cleaner, less realistic speech. We release the data and models to support robust ASR for predominantly oral languages.

Evaluating the Capability of Video Question Generation for Expert Knowledge Elicitation

Skilled human interviewers can extract valuable information from experts. This raises a fundamental question: what makes some questions more effective than others? To address this, a quantitative evaluation of question-generation models is essential. Video question generation (VQG) is a topic for video question answering (VideoQA), where questions are generated for given answers. Their evaluation typically focuses on the ability to answer questions, rather than the quality of generated questions. In contrast, we focus on the question quality in eliciting unseen knowledge from human experts. For a continuous improvement of VQG models, we propose a protocol that evaluates the ability by simulating question-answering communication with experts using a question-to-answer retrieval. We obtain the retriever by constructing a novel dataset, EgoExoAsk, which comprises 27,666 QA pairs generated from Ego-Exo4D's expert commentary annotation. The EgoExoAsk training set is used to obtain the retriever, and the benchmark is constructed on the validation set with Ego-Exo4D video segments. Experimental results demonstrate our metric reasonably aligns with question generation settings: models accessing richer context are evaluated better, supporting that our protocol works as intended. The EgoExoAsk dataset is available in https://github.com/omron-sinicx/VQG4ExpertKnowledge .

Simulstream: Open-Source Toolkit for Evaluation and Demonstration of Streaming Speech-to-Text Translation Systems

Streaming Speech-to-Text Translation (StreamST) requires producing translations concurrently with incoming speech, imposing strict latency constraints and demanding models that balance partial-information decision-making with high translation quality. Research efforts on the topic have so far relied on the SimulEval repository, which is no longer maintained and does not support systems that revise their outputs. In addition, it has been designed for simulating the processing of short segments, rather than long-form audio streams, and it does not provide an easy method to showcase systems in a demo. As a solution, we introduce simulstream, the first open-source framework dedicated to unified evaluation and demonstration of StreamST systems. Designed for long-form speech processing, it supports not only incremental decoding approaches, but also re-translation methods, enabling for their comparison within the same framework both in terms of quality and latency. In addition, it also offers an interactive web interface to demo any system built within the tool.

Large-Language Memorization During the Classification of United States Supreme Court Cases

Large-language models (LLMs) have been shown to respond in a variety of ways for classification tasks outside of question-answering. LLM responses are sometimes called "hallucinations" since the output is not what is ex pected. Memorization strategies in LLMs are being studied in detail, with the goal of understanding how LLMs respond. We perform a deep dive into a classification task based on United States Supreme Court (SCOTUS) decisions. The SCOTUS corpus is an ideal classification task to study for LLM memory accuracy because it presents significant challenges due to extensive sentence length, complex legal terminology, non-standard structure, and domain-specific vocabulary. Experimentation is performed with the latest LLM fine tuning and retrieval-based approaches, such as parameter-efficient fine-tuning, auto-modeling, and others, on two traditional category-based SCOTUS classification tasks: one with 15 labeled topics and another with 279. We show that prompt-based models with memories, such as DeepSeek, can be more robust than previous BERT-based models on both tasks scoring about 2 points better than previous models not based on prompting.

PediatricAnxietyBench: Evaluating Large Language Model Safety Under Parental Anxiety and Pressure in Pediatric Consultations

Large language models (LLMs) are increasingly consulted by parents for pediatric guidance, yet their safety under real-world adversarial pressures is poorly understood. Anxious parents often use urgent language that can compromise model safeguards, potentially causing harmful advice. PediatricAnxietyBench is an open-source benchmark of 300 high-quality queries across 10 pediatric topics (150 patient-derived, 150 adversarial) enabling reproducible evaluation. Two Llama models (70B and 8B) were assessed using a multi-dimensional safety framework covering diagnostic restraint, referral adherence, hedging, and emergency recognition. Adversarial queries incorporated parental pressure patterns, including urgency, economic barriers, and challenges to disclaimers. Mean safety score was 5.50/15 (SD=2.41). The 70B model outperformed the 8B model (6.26 vs 4.95, p<0.001) with lower critical failures (4.8% vs 12.0%, p=0.02). Adversarial queries reduced safety by 8% (p=0.03), with urgency causing the largest drop (-1.40). Vulnerabilities appeared in seizures (33.3% inappropriate diagnosis) and post-vaccination queries. Hedging strongly correlated with safety (r=0.68, p<0.001), while emergency recognition was absent. Model scale influences safety, yet all models showed vulnerabilities to realistic parental pressures. PediatricAnxietyBench provides a reusable adversarial evaluation framework to reveal clinically significant failure modes overlooked by standard benchmarks.

Semantic Tree Inference on Text Corpa using a Nested Density Approach together with Large Language Model Embeddings

Semantic text classification has undergone significant advances in recent years due to the rise of large language models (LLMs) and their high dimensional embeddings. While LLM-embeddings are frequently used to store and retrieve text by semantic similarity in vector databases, the global structure semantic relationships in text corpora often remains opaque. Herein we propose a nested density clustering approach, to infer hierarchical trees of semantically related texts. The method starts by identifying texts of strong semantic similarity as it searches for dense clusters in LLM embedding space. As the density criterion is gradually relaxed, these dense clusters merge into more diffuse clusters, until the whole dataset is represented by a single cluster -- the root of the tree. By embedding dense clusters into increasingly diffuse ones, we construct a tree structure that captures hierarchical semantic relationships among texts. We outline how this approach can be used to classify textual data for abstracts of scientific abstracts as a case study. This enables the data-driven discovery research areas and their subfields without predefined categories. To evaluate the general applicability of the method, we further apply it to established benchmark datasets such as the 20 Newsgroups and IMDB 50k Movie Reviews, demonstrating its robustness across domains. Finally we discuss possible applications on scientometrics, topic evolution, highlighting how nested density trees can reveal semantic structure and evolution in textual datasets.

A stylometric analysis of speaker attribution from speech transcripts

Forensic scientists often need to identify an unknown speaker or writer in cases such as ransom calls, covert recordings, alleged suicide notes, or anonymous online communications, among many others. Speaker recognition in the speech domain usually examines phonetic or acoustic properties of a voice, and these methods can be accurate and robust under certain conditions. However, if a speaker disguises their voice or employs text-to-speech software, vocal properties may no longer be reliable, leaving only their linguistic content available for analysis. Authorship attribution methods traditionally use syntactic, semantic, and related linguistic information to identify writers of written text (authorship attribution). In this paper, we apply a content-based authorship approach to speech that has been transcribed into text, using what a speaker says to attribute speech to individuals (speaker attribution). We introduce a stylometric method, StyloSpeaker, which incorporates character, word, token, sentence, and style features from the stylometric literature on authorship, to assess whether two transcripts were produced by the same speaker. We evaluate this method on two types of transcript formatting: one approximating prescriptive written text with capitalization and punctuation and another normalized style that removes these conventions. The transcripts' conversation topics are also controlled to varying degrees. We find generally higher attribution performance on normalized transcripts, except under the strongest topic control condition, in which overall performance is highest. Finally, we compare this more explainable stylometric model to black-box neural approaches on the same data and investigate which stylistic features most effectively distinguish speakers.

M$^3$KG-RAG: Multi-hop Multimodal Knowledge Graph-enhanced Retrieval-Augmented Generation

Retrieval-Augmented Generation (RAG) has recently been extended to multimodal settings, connecting multimodal large language models (MLLMs) with vast corpora of external knowledge such as multimodal knowledge graphs (MMKGs). Despite their recent success, multimodal RAG in the audio-visual domain remains challenging due to 1) limited modality coverage and multi-hop connectivity of existing MMKGs, and 2) retrieval based solely on similarity in a shared multimodal embedding space, which fails to filter out off-topic or redundant knowledge. To address these limitations, we propose M$^3$KG-RAG, a Multi-hop Multimodal Knowledge Graph-enhanced RAG that retrieves query-aligned audio-visual knowledge from MMKGs, improving reasoning depth and answer faithfulness in MLLMs. Specifically, we devise a lightweight multi-agent pipeline to construct multi-hop MMKG (M$^3$KG), which contains context-enriched triplets of multimodal entities, enabling modality-wise retrieval based on input queries. Furthermore, we introduce GRASP (Grounded Retrieval And Selective Pruning), which ensures precise entity grounding to the query, evaluates answer-supporting relevance, and prunes redundant context to retain only knowledge essential for response generation. Extensive experiments across diverse multimodal benchmarks demonstrate that M$^3$KG-RAG significantly enhances MLLMs' multimodal reasoning and grounding over existing approaches.

Fluent Alignment with Disfluent Judges: Post-training for Lower-resource Languages

We propose a post-training method for lower-resource languages that preserves fluency of language models even when aligned by disfluent reward models. Preference-optimization is now a well-researched topic, but previous work has mostly addressed models for English and Chinese. Lower-resource languages lack both datasets written by native speakers and language models capable of generating fluent synthetic data. Thus, in this work, we focus on developing a fluent preference-aligned language model without any instruction-tuning data in the target language. Our approach uses an on-policy training method, which we compare with two common approaches: supervised finetuning on machine-translated data and multilingual finetuning. We conduct a case study on Norwegian Bokmål and evaluate fluency through native-speaker assessments. The results show that the on-policy aspect is crucial and outperforms the alternatives without relying on any hard-to-obtain data.

Impacts of Racial Bias in Historical Training Data for News AI

AI technologies have rapidly moved into business and research applications that involve large text corpora, including computational journalism research and newsroom settings. These models, trained on extant data from various sources, can be conceptualized as historical artifacts that encode decades-old attitudes and stereotypes. This paper investigates one such example trained on the broadly-used New York Times Annotated Corpus to create a multi-label classifier. Our use in research settings surfaced the concerning "blacks" thematic topic label. Through quantitative and qualitative means we investigate this label's use in the training corpus, what concepts it might be encoding in the trained classifier, and how those concepts impact our model use. Via the application of explainable AI methods, we find that the "blacks" label operates partially as a general "racism detector" across some minoritized groups. However, it performs poorly against expectations on modern examples such as COVID-19 era anti-Asian hate stories, and reporting on the Black Lives Matter movement. This case study of interrogating embedded biases in a model reveals how similar applications in newsroom settings can lead to unexpected outputs that could impact a wide variety of potential uses of any large language model-story discovery, audience targeting, summarization, etc. The fundamental tension this exposes for newsrooms is how to adopt AI-enabled workflow tools while reducing the risk of reproducing historical biases in news coverage.