Transcription and Recognition of Italian Parliamentary Speeches Using Vision-Language Models

Parliamentary proceedings represent a rich yet challenging resource for computational analysis, particularly when preserved only as scanned historical documents. Existing efforts to transcribe Italian parliamentary speeches have relied on traditional Optical Character Recognition pipelines, resulting in transcription errors and limited semantic annotation. In this paper, we propose a pipeline based on Vision-Language Models for the automatic transcription, semantic segmentation, and entity linking of Italian parliamentary speeches. The pipeline employs a specialised OCR model to extract text while preserving reading order, followed by a large-scale Vision-Language Model that performs transcription refinement, element classification, and speaker identification by jointly reasoning over visual layout and textual content. Extracted speakers are then linked to the Chamber of Deputies knowledge base through SPARQL queries and a multi-strategy fuzzy matching procedure. Evaluation against an established benchmark demonstrates substantial improvements both in transcription quality and speaker tagging.

LLM Probe: Evaluating LLMs for Low-Resource Languages

Despite rapid advances in large language models (LLMs), their linguistic abilities in low-resource and morphologically rich languages are still not well understood due to limited annotated resources and the absence of standardized evaluation frameworks. This paper presents LLM Probe, a lexicon-based assessment framework designed to systematically evaluate the linguistic skills of LLMs in low-resource language environments. The framework analyzes models across four areas of language understanding: lexical alignment, part-of-speech recognition, morphosyntactic probing, and translation accuracy. To illustrate the framework, we create a manually annotated benchmark dataset using a low-resource Semitic language as a case study. The dataset comprises bilingual lexicons with linguistic annotations, including part-of-speech tags, grammatical gender, and morphosyntactic features, which demonstrate high inter-annotator agreement to ensure reliable annotations. We test a variety of models, including causal language models and sequence-to-sequence architectures. The results reveal notable differences in performance across various linguistic tasks: sequence-to-sequence models generally excel in morphosyntactic analysis and translation quality, whereas causal models demonstrate strong performance in lexical alignment but exhibit weaker translation accuracy. Our results emphasize the need for linguistically grounded evaluation to better understand LLM limitations in low-resource settings. We release LLM Probe and the accompanying benchmark dataset as open-source tools to promote reproducible benchmarking and to support the development of more inclusive multilingual language technologies.

On the Role of Encoder Depth: Pruning Whisper and LoRA Fine-Tuning in SLAM-ASR

Automatic speech recognition (ASR) has advanced rapidly in recent years, driven by large-scale pretrained models and end-to-end architectures such as SLAM-ASR. A key component of SLAM-ASR systems is the Whisper speech encoder, which provides robust acoustic representations. While model pruning has been explored for the full Whisper encoder-decoder architecture, its impact within the SLAM-ASR setting remains under-investigated. In this work, we analyze the effects of layer pruning in the Whisper encoder when used as the acoustic backbone of SLAM-ASR. We further examine the extent to which LoRA-based fine-tuning can recover performance degradation caused by pruning. Experiments conducted across three Whisper variants (Small, Medium, Large-v2), three languages representing distinct resource levels (Danish, Dutch, English), and over 200 training runs demonstrate that pruning two encoder layers causes only 2-4% WER degradation, and that combining this pruning with LoRA adaptation consistently outperforms the unpruned baseline while reducing total parameters by 7-14%. Moreover, our error analysis reveals that LoRA primarily compensates through the language model's linguistic priors, reducing total word errors by 11-21% for Dutch and English, with substitutions and deletions showing the largest reductions. However, for low-resource Danish, the reduction is smaller (4-7%), and LoRA introduces increased insertion errors, indicating that compensation effectiveness depends on the LLM's pre-existing language proficiency and available training data.

Identification and Anonymization of Named Entities in Unstructured Information Sources for Use in Social Engineering Detection

This study addresses the challenge of creating datasets for cybercrime analysis while complying with the requirements of regulations such as the General Data Protection Regulation (GDPR) and Organic Law 10/1995 of the Penal Code. To this end, a system is proposed for collecting information from the Telegram platform, including text, audio, and images; the implementation of speech-to-text transcription models incorporating signal enhancement techniques; and the evaluation of different Named Entity Recognition (NER) solutions, including Microsoft Presidio and AI models designed using a transformer-based architecture. Experimental results indicate that Parakeet achieves the best performance in audio transcription, while the proposed NER solutions achieve the highest f1-score values in detecting sensitive information. In addition, anonymization metrics are presented that allow evaluation of the preservation of structural coherence in the data, while simultaneously guaranteeing the protection of personal information and supporting cybersecurity research within the current legal framework.

Distilling Conversations: Abstract Compression of Conversational Audio Context for LLM-based ASR

Standard LLM-based speech recognition systems typically process utterances in isolation, limiting their ability to leverage conversational context. In this work, 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 token sequence grows rapidly with conversation length. To address this, we propose Abstract Compression, which replaces the audio portion of prior turns with a fixed number of learned latent tokens while retaining corresponding transcripts explicitly. On both in-domain and out-of-domain test sets, the compressed 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 trade-offs.

Crab: Multi Layer Contrastive Supervision to Improve Speech Emotion Recognition Under Both Acted and Natural Speech Condition

Speech Emotion Recognition (SER) in real-world scenarios remains challenging due to severe class imbalance and the prevalence of spontaneous, natural speech. While recent approaches leverage self-supervised learning (SSL) representations and multimodal fusion of speech and text, most existing methods apply supervision only at the final classification layer, limiting the discriminative power of intermediate representations. In this work, we propose Crab (Contrastive Representation and Multimodal Aligned Bottleneck), a bimodal Cross-Modal Transformer architecture that integrates speech representations from WavLM and textual representations from RoBERTa, together with a novel \textit{Multi Layer Contrastive Supervision} (MLCS) strategy. MLCS injects multi-positive contrastive learning signals at multiple layers of the network, encouraging emotionally discriminative representations throughout the model without introducing additional parameters at inference time. To further address data imbalance, we adopt weighted cross-entropy during training. We evaluate the proposed approach on three benchmark datasets covering different degrees of emotional naturalness: IEMOCAP, MELD, and MSP-Podcast 2.0. Experimental results demonstrate that Crab consistently outperforms strong unimodal and multimodal baselines across all datasets, with particularly large gains under naturalistic and highly imbalanced conditions. These findings highlight the effectiveness of \textit{Multi Layer Contrastive Supervision} as a general and robust strategy for SER. Official implementation can be found in https://github.com/AI-Unicamp/Crab.

How Class Ontology and Data Scale Affect Audio Transfer Learning

Transfer learning is a crucial concept within deep learning that allows artificial neural networks to benefit from a large pre-training data basis when confronted with a task of limited data. Despite its ubiquitous use and clear benefits, there are still many open questions regarding the inner workings of transfer learning and, in particular, regarding the understanding of when and how well it works. To that extent, we perform a rigorous study focusing on audio-to-audio transfer learning, in which we pre-train various model states on (ontology-based) subsets of AudioSet and fine-tune them on three computer audition tasks, namely acoustic scene recognition, bird activity recognition, and speech command recognition. We report that increasing the number of samples and classes in the pre-training data both have a positive impact on transfer learning. This is, however, generally surpassed by similarity between pre-training and the downstream task, which can lead the model to learn comparable features.

Goodness-of-pronunciation without phoneme time alignment

In speech evaluation, an Automatic Speech Recognition (ASR) model often computes time boundaries and phoneme posteriors for input features. However, limited data for ASR training hinders expansion of speech evaluation to low-resource languages. Open-source weakly-supervised models are capable of ASR over many languages, but they are frame-asynchronous and not phonemic, hindering feature extraction for speech evaluation. This paper proposes to overcome incompatibilities for feature extraction with weakly-supervised models, easing expansion of speech evaluation to low-resource languages. Phoneme posteriors are computed by mapping ASR hypotheses to a phoneme confusion network. Word instead of phoneme-level speaking rate and duration are used. Phoneme and frame-level features are combined using a cross-attention architecture, obviating phoneme time alignment. This performs comparably with standard frame-synchronous features on English speechocean762 and low-resource Tamil datasets.

From Oracle to Noisy Context: Mitigating Contextual Exposure Bias in Speech-LLMs

Contextual automatic speech recognition (ASR) with Speech-LLMs is typically trained with oracle conversation history, but relies on error-prone history at inference, causing a train-test mismatch in the context channel that we term contextual exposure bias. We propose a unified training framework to improve robustness under realistic histories: (i) Teacher Error Knowledge by using Whisper large-v3 hypotheses as training-time history, (ii) Context Dropout to regularize over-reliance on history, and (iii) Direct Preference Optimization (DPO) on curated failure cases. Experiments on TED-LIUM 3 (in-domain) and zero-shot LibriSpeech (out-of-domain) show consistent gains under predicted-history decoding. With a two-utterance history as context, SFT with Whisper hypotheses reduce WER from 5.59% (oracle-history training) to 5.47%, and DPO further improves to 5.17%. Under irrelevant-context attacks, DPO yields the smallest degradation (5.17% -> 5.63%), indicating improved robustness to misleading context. Our code and models are published on https://github.com/XYGuo1996/Contextual_Speech_LLMs.

Back to Basics: Revisiting ASR in the Age of Voice Agents

Automatic speech recognition (ASR) systems have achieved near-human accuracy on curated benchmarks, yet still fail in real-world voice agents under conditions that current evaluations do not systematically cover. Without diagnostic tools that isolate specific failure factors, practitioners cannot anticipate which conditions, in which languages, will cause what degree of degradation. We introduce WildASR, a multilingual (four-language) diagnostic benchmark sourced entirely from real human speech that factorizes ASR robustness along three axes: environmental degradation, demographic shift, and linguistic diversity. Evaluating seven widely used ASR systems, we find severe and uneven performance degradation, and model robustness does not transfer across languages or conditions. Critically, models often hallucinate plausible but unspoken content under partial or degraded inputs, creating concrete safety risks for downstream agent behavior. Our results demonstrate that targeted, factor-isolated evaluation is essential for understanding and improving ASR reliability in production systems. Besides the benchmark itself, we also present three analytical tools that practitioners can use to guide deployment decisions.