Speech foundation models have demonstrated exceptional capabilities in speech-related tasks. Nevertheless, these models often struggle with non-verbal audio data, such as vocalizations, baby crying, etc., which are critical for various real-world applications. Audio foundation models well handle non-speech data but also fail to capture the nuanced features of non-verbal human sounds. In this work, we aim to overcome the above shortcoming and propose a novel foundation model, termed voc2vec, specifically designed for non-verbal human data leveraging exclusively open-soruce non-verbal audio datasets. We employ a collection of 10 datasets covering around 125 hours of non-verbal audio. Experimental results prove that voc2vec is effective in non-verbal vocalization classification, and it outperforms conventional speech and audio foundation models. Moreover, voc2vec consistently outperforms strong baselines, namely OpenSmile and emotion2vec, on six different benchmark datasets. To the best of the authors’ knowledge, voc2vec is the first universal representation model for vocalization tasks.

This work aims to tackle the Parkinson’s disease (PD) detection problem from the speech signal in a bilingual setting by proposing an ad-hoc dual-head deep neural architecture for type-based binary classification. One head is specialized for diadochokinetic patterns. The other head looks for natural speech patterns present in continuous spoken utterances. Only one of the two heads is operative accordingly to the nature of the input. Speech representations are extracted from self-supervised learning (SSL) models and wavelet transforms. Adaptive layers, convolutional bottlenecks, and contrastive learning are exploited to reduce variations across languages. Our solution is assessed against two distinct datasets, EWA-DB, and PC-GITA, which cover Slovak and Spanish languages, respectively. Results in-dicate that conventional models trained on a single language dataset struggle with cross-linguistic generalization, and naive combinations of datasets are suboptimal. In contrast, our model improves generalization on both languages, simultaneously.

Voice disorders significantly impact patient quality of life, yet non-invasive automated diagnosis remains under-explored due to both the scarcity of pathological voice data, and the variability in recording sources. This work introduces MVP (Multi-source Voice Pathology detection), a novel approach that leverages transformers operating directly on raw voice signals. We explore three fusion strategies to combine sentence reading and sustained vowel recordings: waveform concatenation, intermediate feature fusion, and decision-level combination. Empirical validation across the German, Portuguese, and Italian languages shows that intermediate feature fusion using transformers best captures the complementary characteristics of both recording types. Our approach achieves up to +13% AUC improvement over single-source methods.

Kolmogorov-Arnold Networks (KANs) have recently emerged as a promising alternative to traditional neural architectures, yet their application to speech processing remains under explored. This work presents the first investigation of KANs for Spoken Language Understanding (SLU) tasks. We experiment with 2D-CNN models on two datasets, integrating KAN layers in five different configurations within the dense block. The best-performing setup, which places a KAN layer between two linear layers, is directly applied to transformer-based models and evaluated on five SLU datasets with increasing complexity. Our results show that KAN layers can effectively replace the linear layers, achieving comparable or superior performance in most cases. Finally, we provide insights into how KAN and linear layers on top of transformers differently attend to input regions of the raw waveforms.

Despite the remarkable progress in end-to-end Automatic Speech Recognition (ASR) engines, accurately transcribing dysarthric speech remains a major challenge. In this work, we proposed a two-stage framework for the Speech Accessibility Project Challenge at INTERSPEECH 2025, which combines cutting-edge speech recognition models with LLM-based generative error correction (GER). We assess different configurations of model scales and training strategies, incorporating specific hypothesis selection to improve transcription accuracy. Experiments on the Speech Accessibility Project dataset demonstrate the strength of our approach on structured and spontaneous speech, while highlighting challenges in single-word recognition. Through comprehensive analysis, we provide insights into the complementary roles of acoustic and linguistic modeling in dysarthric speech recognition.

Formal communications such as public calls, announcements, or regulations are supposed to exhibit respect for diversity in terms of gender, race, age, and disability. However, human writers often lack adequate inclusive writing skills. For instance, they tend to overuse the masculine as a neutral form, mainly because they are self-trained on biased text examples. To overcome this issue, we propose to leverage Generative Artificial Intelligence to support inclusive language writing. Focusing on formal Italian communications, we have designed and developed an AI-assisted tool for non-inclusive text detection and reformulation. Thanks to the joint work with a team of linguistic experts, we first define a set of linguistic criteria necessary to model inclusive writing forms in Italian. Based on these criteria, we collect and annotate a dataset of Italian administrative documents enriched with fine-grained inclusive annotations. Finally, we train deep learning models on the collected data for non-inclusive language detection and inclusive language reformulation tasks. We perform quantitative and human-driven evaluations on the trained models. The best detection model correctly classifies 89% of the sentences, whereas the best reformulation model produces 73% fully correct reformulations. Both models have been integrated into a writing assistance tool acting as a text proofreader and self-learning tool for non-expert writers, namely Inclusively . Once a non-inclusive piece of text is detected, the proposed approach suggests inclusive reformulations. The tool also provides explanations of the models’ outputs to increase system transparency. Furthermore, it allows expert end-users to provide further annotations for system fine-tuning. The trained models and the writing assistance tool are publicly available for research purposes.

Italy exhibits rich linguistic diversity across its territory due to the distinct regional languages spoken in different areas. Recent advances in self-supervised learning provide new opportunities to analyze Italy’s linguistic varieties using speech data alone. This includes the potential to leverage representations learned from large amounts of data to better examine nuances between closely related linguistic varieties. In this study, we focus on automatically identifying the geographic region of origin of speech samples drawn from Italy’s diverse language varieties. We leverage self-supervised learning models to tackle this task and analyze differences and similarities between Italy’s regional languages. In doing so, we also seek to uncover new insights into the relationships among these diverse yet closely related varieties, which may help linguists understand their interconnected evolution and regional development over time and space. To improve the discriminative ability of learned representations, we evaluate several supervised contrastive learning objectives, both as pre-training steps and additional fine-tuning objectives. Experimental evidence shows that pre-trained self-supervised models can effectively identify regions from speech recording. Additionally, incorporating contrastive objectives during fine-tuning improves classification accuracy and yields embeddings that distinctly separate regional varieties, demonstrating the value of combining self-supervised pre-training and contrastive learning for this task.

This work is concerned with devising a robust Parkinson’s (PD) disease detector from speech in real-world operating conditions using (i) foundational models, and (ii) speech enhancement (SE) methods. To this end, we first fine-tune several foundational-based models on the standard PC-GITA (s-PCGITA) clean data. Our results demonstrate superior performance to previously proposed models. Second, we assess the generalization capability of the PD models on the extended PCGITA (e-PC-GITA) recordings, collected in real-world operative conditions, and observe a severe drop in performance moving from ideal to real-world conditions. Third, we align training and testing conditions applaying off-the-shelf SE techniques on e-PC-GITA, and a significant boost in performance is observed only for the foundational-based models. Finally, combining the two best foundational-based models trained on s-PCGITA, namely WavLM Base and Hubert Base, yielded top performance on the enhanced e-PC-GITA

Self-supervised representation learning (SSL) has attained SOTA results on several downstream speech tasks, but SSL-based speech enhancement (SE) solutions still lag behind. To address this issue, we exploit three main ideas: (i) Transformer-based masking generation, (ii) consistency-preserving loss, and (iii) perceptual contrast stretching (PCS). In detail, conformer layers, leveraging an attention mechanism, are introduced to effectively model frame-level representations and obtain the Ideal Ratio Mask (IRM) for SE. Moreover, we incorporate consistency in the loss function, which processes the input to account for the inconsistency effects of signal reconstruction from the spectrogram. Finally, PCS is employed to improve the contrast of input and target features according to perceptual importance. Evaluated on the VoiceBank-DEMAND task, the proposed solution outperforms previously SSL-based SE solutions when tested on several objective metrics, attaining a SOTA PESQ score of 3.54.

This work aims to investigate the use of a recently proposed, attention-free, scalable state-space model (SSM), Mamba, for the speech enhancement (SE) task. In particular, we employ Mamba to deploy different regression-based SE models (SEMamba) with different configurations, namely basic, advanced, causal, and non-causal. Furthermore, loss functions either based on signal-level distances or metric-oriented are considered. Experimental evidence shows that SEMamba attains a competitive PESQ of 3.55 on the VoiceBank-DEMAND dataset with the advanced, non-causal configuration. A new state-of-the-art PESQ of 3.69 is also reported when SEMamba is combined with Perceptual Contrast Stretching (PCS). Compared against Transformed-based equivalent SE solutions, a noticeable FLOPs reduction up to $\sim 12 \%$ is observed with the advanced non-causal configurations. Finally, SEMamba can be used as a pre-processing step before automatic speech recognition (ASR), showing competitive performance against recent SE solutions.

In this paper, we present an encoder-decoder model leveraging Flan-T5 for post-Automatic Speech Recognition (ASR) Generative Speech Error Correction (GenSEC), and we refer to it as FlanEC. We explore its application within the GenSEC framework to enhance ASR outputs by mapping n-best hypotheses into a single output sentence. By utilizing n-best lists from ASR models, we aim to improve the linguistic correctness, accuracy, and grammaticality of final ASR transcriptions. Specifically, we investigate whether scaling the training data and incorporating diverse datasets can lead to significant improvements in post-ASR error correction. We evaluate FlanEC using the HyPoradise dataset, providing a comprehensive analysis of the model’s effectiveness in this domain. Furthermore, we assess the proposed approach under different settings to evaluate model scalability and efficiency, offering valuable insights into the potential of instruction-tuned encoder-decoder models for this task.

Text Style Transfer (TST) is a relevant branch of natural language processing that aims to control the style attributes of a piece of text while preserving its original content. To address TST in the absence of parallel data, Cycle-consistent Generative Adversarial Networks (CycleGANs) have recently emerged as promising solutions. Existing CycleGAN-based TST approaches suffer from the following limitations: (1) They apply self-supervision, based on the cycle-consistency principle, in the latent space. This approach turns out to be less robust to mixed-style inputs, i.e., when the source text is partly in the original and partly in the target style; (2) Generators and discriminators rely on recurrent networks, which are exposed to known issues with long-term text dependencies; (3) The target style is weakly enforced, as the discriminator distinguishes real from fake sentences without explicitly accounting for the generated text's style. We propose a new CycleGAN-based TST approach that applies self-supervision directly at the sequence level to effectively handle mixed-style inputs and employs Transformers to leverage the attention mechanism for both text encoding and decoding. We also employ a pre-trained style classifier to guide the generation of text in the target style while maintaining the original content's meaning. The experimental results achieved on the formality and sentiment transfer tasks show that our approach outperforms existing ones, both CycleGAN-based and not (including an open-source Large Language Model), on benchmark data and shows better robustness to mixed-style inputs.

Recent large-scale Spoken Language Understanding datasets focus predominantly on English and do not account for language-specific phenomena such as particular phonemes or words in different lects.We introduce ITALIC, the first largescale speech dataset designed for intent classification in Italian.The dataset comprises 16,521 crowdsourced audio samples recorded by 70 speakers from various Italian regions and annotated with intent labels and additional metadata.We explore the versatility of ITALIC by evaluating current state-of-the-art speech and text models.Results on intent classification suggest that increasing scale and running language adaptation yield better speech models, monolingual text models outscore multilingual ones, and that speech recognition on ITALIC is more challenging than on existing Italian benchmarks.We release both the dataset and the annotation scheme to streamline the development of new Italian SLU models and language-specific datasets.

Learning environments foster the exchange of large amounts of data among learners and teachers. Summarization techniques leverage information retrieval and machine learning techniques to condense the key information hidden in large data collections into actionable summaries. Their integration into existing education technology systems is particularly appealing as it enables smart, automated solutions to challenging learning tasks such as content curation, accessibility, and personalization. This paper presents a general-purpose summarization-based methodology to learn. It aims at extending the current ed-ucational learning systems by envisaging the integration of summarization methods at different learning stages. Specifically, it tailors the output summaries to different end-users (either teachers or learners), content types (e.g., text, audio, video), and learning goals. With the goal of making the devised methodology actionable, the paper also examines the current role of summa-rization in the learning process and highlights the open directions and perspectives.

Mosquitoes are a major global health problem. They are responsible for the transmission of diseases and can have a large impact on local economies. Monitoring mosquitoes is therefore helpful in preventing the outbreak of mosquito-borne diseases. In this paper, we propose a novel data-driven approach that leverages Transformer-based models for the identification of mosquitoes in audio recordings. The task aims at detecting the time intervals corresponding to the acoustic mosquito events in an audio signal. We formulate the problem as a sequence tagging task and train a Transformer-based model using a real-world dataset collecting mosquito recordings. By leveraging the sequential nature of mosquito recordings, we formulate the training objective so that the input recordings do not require fine-grained annotations. We show that our approach is able to outperform baseline methods using standard evaluation metrics, albeit suffering from unexpectedly high false negatives detection rates. In view of the achieved results, we propose future directions for the design of more effective mosquito detection models.

Preserving diversity and inclusion is becoming a compelling need in both industry and academia. The ability to use appropriate forms of writing, speaking, and gestures is not widespread even in formal communications such as public calls, public announcements, official reports, and legal documents. The improper use of linguistic expressions can foment unacceptable forms of exclusion, stereotypes as well as forms of verbal violence against minorities, including women. Furthermore, existing machine translation tools are not designed to generate inclusive content.The present paper investigates a joint effort of the research communities of linguistics and Deep Learning Natural Language Understanding in fighting against non-inclusive, prejudiced language forms. It presents a methodology aimed at tackling the improper use of language in formal communication, with a particular attention paid to Romanic languages (Italian, in particular). State-of-the-art Deep Language Modeling architectures are exploited to automatically identify non-inclusive text snippets, suggest alternative forms, and produce inclusive text rephrasing. A preliminary evaluation conducted on a benchmark dataset shows promising results, i.e., 85% accuracy in predicting inclusive/non-inclusive communications.

Disseminating the main research findings is one of the main requirements to become a successful researcher. Presentation slides are the most common way to present paper content. To support researchers in slide preparation, the NLP research community has explored the use of summarization techniques to automatically generate a draft of the slides consisting of the most salient sentences or phrases. State-of-the-art methods adopt a supervised approach, which first estimates global content relevance using a set of training papers and slides, then performs content selection by optimizing also section-level coverage. However, in several domains and contexts there is a lack of training data, which hinders the use of supervised models. This paper focuses on addressing the above issue by applying unsupervised summarization methods. They are exploited to generate sentence-level summaries of the paper sections, which are then refined by applying an optimization step. Furthermore, it evaluates the quality of the output slides by taking into account the original paper structure as well. The results, achieved on a benchmark collection of papers and slides, show that unsupervised models performed better than supervised ones on specific paper facets, whereas they were competitive in terms of overall quality score.

Timeline summarization aims at presenting long news stories in a compact manner. State-of-the-art approaches first select the most relevant dates from the original event timeline then produce per-date news summaries. Date selection is driven by either per-date news content or date-level references. When coping with complex event data, characterized by inherent news flow redundancy, this pipeline may encounter relevant issues in both date selection and summarization due to a limited use of news content in date selection and no use of high-level temporal references (e.g., the past month). This paper proposes a paradigm shift in timeline summarization aimed at overcoming the above issues. It presents a new approach, namely Summarize Date First, which focuses on first generating date-level summaries then selecting the most relevant dates on top of summarized knowledge. In the latter stage, it performs date aggregations to consider high-level temporal references as well. The proposed pipeline also supports frequent incremental timeline updates more efficiently than previous approaches. We tested our unsupervised approach both on existing benchmark datasets and on a newly proposed benchmark dataset describing the COVID-19 news timeline. The achieved results were superior to state-of-the-art unsupervised methods and competitive against supervised ones.