论文检索

Yejin Jeon, Youngjae Kim, Jihyun Lee, Hyounghun Kim, Gary Lee

For individuals who have experienced traumatic events such as strokes, speech may no longer be a viable means of communication. While text-to-speech (TTS) can be used as a communication aid since it generates synthetic speech, it fails to preserve the user’s own voice. As such, face-to-voice (FTV) synthesis, which derives corresponding voices from facial images, provides a promising alternative. However, existing methods rely on pre-trained visual encoders, and finetune them to align with speech embeddings, which strips fine-grained information from facial inputs such as gender or ethnicity, despite their known correlation with vocal traits. Moreover, these pipelines are multi-stage, which requires separate training of multiple components, thus leading to training inefficiency. To address these limitations, we utilize fine-grained facial attribute modeling by decomposing facial images into non-overlapping segments and progressively integrating them into a multi-granular representation. This representation is further refined through multi-task learning of speaker attributes such as gender and ethnicity at both the visual and acoustic domains. Moreover, to improve alignment robustness, we adopt a multi-view training strategy by pairing various visual perspectives of a speaker in terms of different angles and lighting conditions, with identical speech recordings. Extensive subjective and objective evaluations confirm that our approach substantially enhances face-voice congruence and synthesis stability.

Weiting Tan, Jiachen Lian, Hirofumi Inaguma, Paden Tomasello, Philipp Koehn, Xutai Ma

We present an Audio-Visual Language Model (AVLM) for expressive speech generation by integrating full-face visual cues into a pre-trained expressive speech model. We explore multiple visual encoders and multimodal fusion strategies during pre-training to identify the most effective integration approach. Subsequent fine-tuning on emotion recognition and expressive dialogue tasks yields substantial gains over speech-only baselines (e.g., +5 F1 in emotion recognition). AVLM highlights the value of expressive visual information in guiding speech generation and offers a foundation for end-to-end multimodal conversational systems.

Cheng-Han Chiang, Xiaofei Wang, Chung-Ching Lin, Kevin Lin, Linjie Li, Radu Kopetz, Yao Qian, Zhendong Wang, Zhengyuan Yang, Hung-yi Lee 等

Audio-aware large language models (ALLMs) can understand the textual and non-textual information in the audio input. In this paper, we explore using ALLMs as an automatic judge to assess the speaking styles of speeches. We use ALLM judges to evaluate the speeches generated by SLMs on two tasks: voice style instruction following and role-playing. The speaking style we consider includes emotion, volume, speaking pace, word emphasis, pitch control, and non-verbal elements. We use four spoken language models (SLMs) to complete the two tasks and use humans and ALLMs to judge the SLMs’ responses. We compare two ALLM judges, GPT-4o-audio and Gemini-2.5-pro, with human evaluation results and show that the agreement between Gemini and human judges is comparable to the agreement between human evaluators. These promising results show that ALLMs can be used as a judge to evaluate SLMs. Our results also reveal that current SLMs, even GPT-4o-audio, still have room for improvement in controlling the speaking style and generating natural dialogues.

Linyang He, Qiaolin Wang, Xilin Jiang, Nima Mesgarani

Transformer-based speech language models (SLMs) have significantly improved neural speech recognition and understanding. While existing research has examined how well SLMs encode shallow acoustic and phonetic features, the extent to which SLMs encode nuanced syntactic and conceptual features remains unclear. By drawing parallels with linguistic competence assessments for large language models, this study is the first to systematically evaluate the presence of contextual syntactic and semantic features across SLMs for self-supervised learning (S3M), automatic speech recognition (ASR), speech compression (codec), and as the encoder for auditory large language models (AudioLLMs). Through minimal pair designs and diagnostic feature analysis across 71 tasks spanning diverse linguistic levels, our layer-wise and time-resolved analysis uncovers that 1) all speech encode grammatical features more robustly than conceptual ones. 2) Despite never seeing text, S3M match or surpass ASR encoders on every linguistic level, demonstrating that rich grammatical and even conceptual knowledge can arise purely from audio. 3) S3M representations peak mid-network and then crash in the final layers, whereas ASR and AudioLLM encoders maintain or improve, reflecting how pre-training objectives reshape late-layer content. 4) Temporal probing further shows that S3Ms encode grammatical cues 500 ms before a word begins, whereas AudioLLMs distribute evidence more evenly—indicating that objectives shape not only where but also when linguistic information is most salient. Together, these findings establish the first large-scale map of contextual syntax and semantics in speech models and highlight both the promise and the limits of current SLM training paradigms.

Alexandre Costa Ferro Filho, Rafaello Virgilli, Lucas Alcantara Souza, F S de Oliveira, Marcelo Henrique Lopes Ferreira, Daniel Tunnermann, Gustavo Dos Reis Oliveira, Anderson Da Silva Soares, Arlindo Rodrigues Galvão Filho

The detection of audio deepfakes (ADD) has become increasingly important due to the rapid evolution of generative speech models. However, progress in this field remains uneven across languages, particularly for low-resource languages like Portuguese, which lack high-quality datasets. In this paper, we introduce BRSpeech-DF, the first publicly available ADD dataset for Portuguese, encompassing both Brazilian and European variants. The dataset contains over 458,000 utterances, including a smaller portion of real speech from 62 speakers and a large collection of synthetic samples generated using multiple zero-shot text-to-speech (TTS) models, each conditioned on the original speaker’s voice. By providing this resource, our objective is to support the development of robust, multilingual detection systems, thereby advancing equity in speech forensics and security research. BRSpeech-DF addresses a significant gap in annotated data for underrepresented languages, facilitating more inclusive and generalizable advancements in synthetic speech detection.

Fagun Patel, Duc Q. Nguyen, Sang T. Truong, Jody Vaynshtok, Sanmi Koyejo, Nick Haber

According to the U.S. National Institutes of Health, more than 3.4 million children experience speech disorders that require clinical intervention. The number of speech-language pathologists (SLPs) is roughly 20 times fewer than the number of affected children, highlighting a significant gap in children’s care and a pressing need for technological support that improves the productivity of SLPs. State-of-the-art multimodal language models (MLMs) show promise for supporting SLPs, but their use remains underexplored largely due to a limited understanding of their performance in high-stakes clinical settings. To address this gap, we collaborate with domain experts to develop a taxonomy of real-world use cases of MLMs in speech-language pathologies. Building on this taxonomy, we introduce the first comprehensive benchmark for evaluating MLM across five core use cases, each containing 1,000 manually annotated data points. This benchmark includes robustness and sensitivity tests under various settings, including background noise, speaker gender, and accent. Our evaluation of 15 state-of-the-art MLMs reveals that no single model consistently outperforms others across all tasks. Notably, we find systematic disparities, with models performing better on male speakers, and observe that chain-of-thought prompting can degrade performance on classification tasks with large label spaces and narrow decision boundaries. Furthermore, we study fine-tuning MLMs on domain-specific data, achieving improvements of over 30% compared to base models. These findings highlight both the potential and limitations of current MLMs for speech-language pathology applications, underscoring the need for further research and targeted development.

Padakanti Srijith, Khushbu Pahwa, Radhika Mamidi, Bapi Raju Surampudi, Manish Gupta, Subba Reddy Oota

Although speech language models are expected to align well with brain language processing during speech comprehension, recent studies have found that they fail to capture brain-relevant semantics beyond low-level features. Surprisingly, text-based language models exhibit stronger alignment with brain language regions, as they better capture brain-relevant semantics. However, no prior work has examined the alignment effectiveness of text/speech representations from multimodal models. This raises several key questions: Can speech embeddings from such multimodal models capture brain-relevant semantics through cross-modal interactions? Which modality can take advantage of this synergistic multimodal understanding to improve alignment with brain language processing? Can text/speech representations from such multimodal models outperform unimodal models? To address these questions, we systematically analyze multiple multimodal models, extracting both text- and speech-based representations to assess their alignment with MEG brain recordings during naturalistic story listening. We find that text embeddings from both multimodal and unimodal models significantly outperform speech embeddings from these models. Specifically, multimodal text embeddings exhibit a peak around 200 ms, suggesting that they benefit from speech embeddings, with heightened activity during this time period. However, speech embeddings from these multimodal models still show a similar alignment compared to their unimodal counterparts, suggesting that they do not gain meaningful semantic benefits over text-based representations. These results highlight an asymmetry in cross-modal knowledge transfer, where the text modality benefits more from speech information, but not vice versa.

Seung-Bin Kim, Jun-Hyeok Cha, Hyung-Seok Oh, Heejin Choi, Seong-Whan Lee

Recent advancements in speech synthesis have significantly improved the audio quality and pronunciation of synthesized speech. To further advance toward human-like conversational speech synthesis, this paper presents FillerSpeech, a novel speech synthesis framework that enables natural filler insertion and control over filler style. To address this, we construct a filler-inclusive speech data, derived from the open-source large-scale speech corpus. This data includes fillers with pitch and duration information. For the generation and style control of natural fillers, we propose a method that tokenizes the filler style and utilizes cross-attention with the input text. Furthermore, we introduce a large language model-based filler prediction method that enables natural insertion of fillers even when only text input is provided. The experimental results demonstrate that the constructed dataset is valid and that our proposed methods for filler style control and filler prediction are effective.

Emanuele Moscato, Tiancheng Hu, Matthias Orlikowski, Paul Röttger, Debora Nozza

Personalized content moderation can protect users from harm while facilitating free expression by tailoring moderation decisions to individual preferences rather than enforcing universal rules. However, content moderation that is fully personalized to individual preferences, no matter what these preferences are, may lead to even the most hazardous types of content being propagated on social media. In this paper, we explore this risk using hate speech as a case study. Certain types of hate speech are illegal in many countries. We show that, while fully personalized hate speech detection models increase overall user welfare (as measured by user-level classification performance), they also make predictions that violate such legal hate speech boundaries, especially when tailored to users who tolerate highly hateful content. To address this problem, we enforce legal boundaries in personalized hate speech detection by overriding predictions from personalized models with those from a boundary classifier. This approach significantly reduces legal violations while minimally affecting overall user welfare. Our findings highlight both the promise and the risks of personalized moderation, and offer a practical solution to balance user preferences with legal and ethical obligations.

Vishnu Raja, Adithya V Ganesan, Anand Syamkumar, Ritwik Banerjee, H. Schwartz

State-of-the-art automatic speech recognition (ASR) models like Whisper perform poorly on atypical speech, such as that produced by individuals with dysarthria. Past works for atypical speech have mostly investigated fully personalized (or idiosyncratic) models, but modeling strategies that can both generalize and handle idiosyncrasy could be more effective for capturing atypical speech. To investigate this, we compare four strategies: (a) *normative* models trained on typical speech (no personalization), (b) *idiosyncratic* models completely personalized to individuals, (c) *dysarthric-normative* models trained on other dysarthric speakers, and (d) *dysarthric-idiosyncratic* models which combine strategies by first modeling normative patterns before adapting to individual speech. In this case study, we find the dysarthric-idiosyncratic model performs better than the idiosyncratic approach while requiring less than half as much personalized data (36.43 WER with 128 train size vs. 36.99 with 256). Further, we found that tuning the speech encoder alone (as opposed to the LM decoder) yielded the best results, reducing word error rate from 71% to 32% on average. Our findings highlight the value of leveraging both normative (cross-speaker) and idiosyncratic (speaker-specific) patterns to improve ASR for underrepresented speech populations. [GitHub: VishnuRaja98/Dysarthric-Speech-Transcription](https://github.com/VishnuRaja98/Dysarthric-Speech-Transcription)

Siddhant Bikram Shah, Kristina T. Johnson

Nonverbal vocalizations are an essential component of human communication, conveying rich information without linguistic content. However, their computational analysis is hindered by a lack of lexical anchors in the data, compounded by biased and imbalanced data distributions. While disentangled representation learning has shown promise in isolating specific speech features, its application to nonverbal vocalizations remains unexplored. In this paper, we introduce N-CORE, a novel backbone-agnostic framework designed to disentangle intertwined features like emotion and speaker information from nonverbal vocalizations by leveraging N views of audio samples to learn invariance to specific transformations. N-CORE achieves competitive performance compared to state-of-the-art methods for emotion and speaker classification on the VIVAE, ReCANVo, and ReCANVo-Balanced datasets. We further propose an emotion perturbation function that disrupts affective information while preserving speaker information in audio signals for emotion-invariant speaker classification. Our work informs research directions on paralinguistic speech processing, including clinical diagnoses of atypical speech and longitudinal analysis of communicative development. Our code is available at https://github.com/SiddhantBikram/N-CORE.

Finlay Fehlauer, Kyle Mahowald, Tiago Pimentel

In this paper, we investigate the convergence of language models (LMs) trained under different random seeds, measuring convergence as the expected per-token Kullback–Leibler (KL) divergence across seeds. By comparing LM convergence as a function of model size and training checkpoint, we identify a four-phase convergence pattern: (i) an initial \textbf{uniform phase}, (ii) a \textbf{sharp-convergence phase}, (iii) a \textbf{sharp-divergence phase}, and (iv) a \textbf{slow-reconvergence phase}. Further, we observe that larger models reconverge faster in later training stages, while smaller models never actually reconverge; these results suggest that a certain model size may be necessary to learn stable distributions. Restricting our analysis to specific token frequencies, or part-of-speech (PoS) tags further reveals that convergence is uneven across linguistic categories: frequent tokens and function words converge faster and more reliably than their counterparts (infrequent tokens and content words). Overall, our findings highlight factors that influence the stability of the learned distributions in model training.

Yongjian Chen, Antonio Toral

Neural text-based models for detecting machine-translated texts can rely on named entities (NEs) as classification shortcuts. While masking NEs encourages learning genuine translationese signals, it degrades the classification performance. Incorporating speech features compensates for this loss, but their interaction with NE reliance requires careful investigation. Through systematic attribution analysis across modalities, we find that bimodal integration leads to more balanced feature utilization, reducing the reliance on NEs in text while moderating overemphasis attribution patterns in speech features.

Srikant Panda, Amit Agarwal, Hitesh Laxmichand Patel

Large Language Models (LLMs) are increasingly deployed across diverse domains but often exhibit disparities in how they handle real life queries. To systematically investigate these effects with various disability context, we introduce AccessEval, a large-scale benchmark evaluating total 21 close & open source LLMs across six real-world domains and nine disability types using paired Neutral and Disability-Aware Queries. We evaluated model outputs with metrics for factual accuracy, sentiment, and social perception.Our analysis reveals that responses to disability-aware queries tend to have higher factual error, more negative tone, and increased stereotyping with social perception compared to neutral queries. These effects show notable variation by domain and disability type. Disabilities affecting hearing, speech and mobility are disproportionately impacted. These disparities reveal persistent forms of ableism, highlighting the need for more comprehensive and nuanced assessment.We further argue that framing bias in terms of model performance within real-world decision making helps to better link model behaviors to the potential harms users may face. This approach guides the development of more effective and tailored fairness interventions. AccessEval, therefore, serves as a crucial tool for advancing equitable and inclusive language technologies.

Abteen Ebrahimi, Adam Wiemerslage, Katharina von der Wense

We present NN-Rank, an algorithm for ranking source languages for cross-lingual transfer, which leverages hidden representations from multilingual models and unlabeled target-language data. We experiment with two pretrained multilingual models and two tasks: part-of-speech tagging (POS) and named entity recognition (NER). We consider 51 source languages and evaluate on 56 and 72 target languages for POS and NER, respectively. When using in-domain data, NN-Rank beats state-of-the-art baselines that leverage lexical and linguistic features, with average improvements of up to 35.56 NDCG for POS and 18.14 NDCG for NER. As prior approaches can fall back to language-level features if target language data is not available, we show that NN-Rank remains competitive using only the Bible, an out-of-domain corpus available for a large number of languages. Ablations on the amount of unlabeled target data show that, for subsets consisting of as few as 25 examples, NN-Rank produces high-quality rankings which achieve 92.8% of the NDCG achieved using all available target data for ranking.

Hauke Licht, Rupak Sarkar, Patrick Y. Wu, Pranav Goel, Niklas Stoehr, Elliott Ash, Alexander Miserlis Hoyle

Many constructs that characterize language, like its complexity or emotionality, have a naturally continuous semantic structure; a public speech is not just “simple” or “complex”, but exists on a continuum between extremes. Although large language models (LLMs) are an attractive tool for measuring scalar constructs, their idiosyncratic treatment of numerical outputs raises questions of how to best apply them. We address these questions with a comprehensive evaluation of LLM-based approaches to scalar construct measurement in social science. Using multiple datasets sourced from the political science literature, we evaluate four approaches: unweighted direct pointwise scoring, aggregation of pairwise comparisons, token-probability-weighted pointwise scoring, and finetuning. Our study finds that pairwise comparisons made by LLMs produce better measurements than simply prompting the LLM to directly output the scores, which suffers from bunching around arbitrary numbers. However, taking the weighted mean over the token probability of scores further improves the measurements over the two previous approaches. Finally, finetuning smaller models with as few as 1,000 training pairs can match or exceed the performance of prompted LLMs.

Anthony Dubreuil, Antoine Gourru, Christine Largeron, Amine Trabelsi

Large Language Models inherit stereotypes from their pretraining data, leading to biased behavior toward certain social groups in many Natural Language Processing tasks, such as hateful speech detection or sentiment analysis. Surprisingly, the evaluation of this kind of bias in stance detection methods has been largely overlooked by the community. Stance Detection involves labeling a statement as being against, in favor, or neutral towards a specific target and is among the most sensitive NLP tasks, as it often relates to political leanings. In this paper, we focus on the bias of Large Language Models when performing stance detection in a zero-shot setting. We automatically annotate posts in pre-existing stance detection datasets with two attributes: dialect or vernacular of a specific group and text complexity/readability, to investigate whether these attributes influence the model’s stance detection decisions. Our results show that LLMs exhibit significant stereotypes in stance detection tasks, such as incorrectly associating pro-marijuana views with low text complexity and African American dialect with opposition to Donald Trump.

Jacob Lee Suchardt, Hana El-Shazli, Pierluigi Cassotti

This paper explores the use of existing state-of-the-art speech recognition models (ASR) for the task of generating narrow phonetic transcriptions using the International Phonetic Alphabet (STIPA). Unlike conventional ASR systems focused on orthographic output for high-resource languages, STIPA can be used as a language-agnostic interface valuable for documenting under-resourced and unwritten languages. We introduce a new dataset for South Levantine Arabic and present the first large-scale evaluation of STIPA models across 51 language families. Additionally, we provide a use case on Sanna, a severely endangered language. Our findings show that fine-tuned ASR models can produce accurate IPA transcriptions with limited supervision, significantly reducing phonetic error rates even in extremely low-resource settings. The results highlight the potential of STIPA for scalable language documentation.

Siyu Liang, Nicolas Ballier, Gina-Anne Levow, Richard Wright

While large multilingual automatic speech recognition (ASR) models achieve remarkable performance, the internal mechanisms of the end-to-end pipeline, particularly concerning fairness and efficacy across languages, remain underexplored. This paper introduces a fine-grained analysis of Whisper’s multilingual decoder, examining its sub-token hypotheses during transcription across languages with various resource levels. Our method traces the beam search path, capturing sub-token guesses and their associated probabilities. Results reveal that higher resource languages benefit from higher likelihood of the correct token being top-ranked, greater confidence, lower predictive entropy, and more diverse alternative candidates. Lower resource languages fare worse on these metrics, but also exhibit distinct clustering patterns in sub-token usage sometimes influenced by typology in our PCA and t-SNE analysis. This sub-token probing uncovers systematic decoding disparities masked by aggregate error rates and points towards targeted interventions to ameliorate the imbalanced development of speech technology.

Mohamed Bayan Kmainasi, Abul Hasnat, Md Arid Hasan, Ali Ezzat Shahroor, Firoj Alam

The proliferation of multimodal content on social media presents significant challenges in understanding and moderating complex, context-dependent issues such as misinformation, hate speech, and propaganda. While efforts have been made to develop resources and propose new methods for automatic detection, limited attention has been given to label detection and the generation of explanation-based rationales for predicted labels. To address this challenge, we introduce MemeXplain, an explanation-enhanced dataset for propaganda memes in Arabic and hateful memes in English, making it the first large-scale resource for these tasks. To solve these tasks, we propose a novel multi-stage optimization approach and train Vision-Language Models (VLMs). Our results demonstrate that this approach significantly improves performance over the base model for both label detection and explanation generation, outperforming the current state-of-the-art with an absolute improvement of approximately 3% on ArMeme and 7% on Hateful Memes. For reproducibility and future research, we aim to make the MemeXplain dataset and scripts publicly available.