BUT Systems for Environmental Sound Deepfake Detection in the ESDD 2026 Challenge

This paper describes the BUT submission to the ESDD 2026 Challenge, specifically focusing on Track 1: Environmental Sound Deepfake Detection with Unseen Generators. To address the critical challenge of generalizing to audio generated by unseen synthesis algorithms, we propose a robust ensemble framework leveraging diverse Self-Supervised Learning (SSL) models. We conduct a comprehensive analysis of general audio SSL models (including BEATs, EAT, and Dasheng) and speech-specific SSLs. These front-ends are coupled with a lightweight Multi-Head Factorized Attention (MHFA) back-end to capture discriminative representations. Furthermore, we introduce a feature domain augmentation strategy based on distribution uncertainty modeling to enhance model robustness against unseen spectral distortions. All models are trained exclusively on the official EnvSDD data, without using any external resources. Experimental results demonstrate the effectiveness of our approach: our best single system achieved Equal Error Rates (EER) of 0.00\%, 4.60\%, and 4.80\% on the Development, Progress (Track 1), and Final Evaluation sets, respectively. The fusion system further improved generalization, yielding EERs of 0.00\%, 3.52\%, and 4.38\% across the same partitions.

LG Uplus System with Multi-Speaker IDs and Discriminator-based Sub-Judges for the WildSpoof Challenge

This paper describes our submission to the WildSpoof Challenge Track 2, which focuses on spoof-aware speaker verification (SASV) in the presence of high-quality text-to-speech (TTS) attacks. We adopt a ResNet-221 back-bone and study two speaker-labeling strategies, namelyDual-Speaker IDs and Multi-Speaker IDs, to explicitly enlarge the margin between bona fide and generated speech in the embedding space. In addition, we propose discriminator-based sub-judge systems that reuse internal features from HiFi-GAN and BigVGAN discriminators, aggregated via multi-query multi-head attentive statistics pooling(MQMHA). Experimental results on the SpoofCeleb corpus show that our system design is effective in improving agnostic detection cost function (a-DCF).

Efficient ASR for Low-Resource Languages: Leveraging Cross-Lingual Unlabeled Data

Automatic speech recognition for low-resource languages remains fundamentally constrained by the scarcity of labeled data and computational resources required by state-of-the-art models. We present a systematic investigation into cross-lingual continuous pretraining for low-resource languages, using Perso-Arabic languages (Persian, Arabic, and Urdu) as our primary case study. Our approach demonstrates that strategic utilization of unlabeled speech data can effectively bridge the resource gap without sacrificing recognition accuracy. We construct a 3,000-hour multilingual corpus through a scalable unlabeled data collection pipeline and employ targeted continual pretraining combined with morphologically-aware tokenization to develop a 300M parameter model that achieves performance comparable to systems 5 times larger. Our model outperforms Whisper Large v3 (1.5B parameters) on Persian and achieves competitive results on Arabic and Urdu despite using significantly fewer parameters and substantially less labeled data. These findings challenge the prevailing assumption that ASR quality scales primarily with model size, revealing instead that data relevance and strategic pretraining are more critical factors for low-resource scenarios. This work provides a practical pathway toward inclusive speech technology, enabling effective ASR for underrepresented languages without dependence on massive computational infrastructure or proprietary datasets.

TeluguST-46: A Benchmark Corpus and Comprehensive Evaluation for Telugu-English Speech Translation

Despite Telugu being spoken by over 80 million people, speech translation research for this morphologically rich language remains severely underexplored. We address this gap by developing a high-quality Telugu--English speech translation benchmark from 46 hours of manually verified CSTD corpus data (30h/8h/8h train/dev/test split). Our systematic comparison of cascaded versus end-to-end architectures shows that while IndicWhisper + IndicMT achieves the highest performance due to extensive Telugu-specific training data, finetuned SeamlessM4T models demonstrate remarkable competitiveness despite using significantly less Telugu-specific training data. This finding suggests that with careful hyperparameter tuning and sufficient parallel data (potentially less than 100 hours), end-to-end systems can achieve performance comparable to cascaded approaches in low-resource settings. Our metric reliability study evaluating BLEU, METEOR, ChrF++, ROUGE-L, TER, and BERTScore against human judgments reveals that traditional metrics provide better quality discrimination than BERTScore for Telugu--English translation. The work delivers three key contributions: a reproducible Telugu--English benchmark, empirical evidence of competitive end-to-end performance potential in low-resource scenarios, and practical guidance for automatic evaluation in morphologically complex language pairs.

Beyond Unified Models: A Service-Oriented Approach to Low Latency, Context Aware Phonemization for Real Time TTS

Lightweight, real-time text-to-speech systems are crucial for accessibility. However, the most efficient TTS models often rely on lightweight phonemizers that struggle with context-dependent challenges. In contrast, more advanced phonemizers with a deeper linguistic understanding typically incur high computational costs, which prevents real-time performance. This paper examines the trade-off between phonemization quality and inference speed in G2P-aided TTS systems, introducing a practical framework to bridge this gap. We propose lightweight strategies for context-aware phonemization and a service-oriented TTS architecture that executes these modules as independent services. This design decouples heavy context-aware components from the core TTS engine, effectively breaking the latency barrier and enabling real-time use of high-quality phonemization models. Experimental results confirm that the proposed system improves pronunciation soundness and linguistic accuracy while maintaining real-time responsiveness, making it well-suited for offline and end-device TTS applications.

A Simple Method to Enhance Pre-trained Language Models with Speech Tokens for Classification

This paper presents a simple method that allows to easily enhance textual pre-trained large language models with speech information, when fine-tuned for a specific classification task. A classical issue with the fusion of many embeddings from audio with text is the large length of the audio sequence compared to the text one. Our method benefits from an existing speech tokenizer trained for Audio Speech Recognition that output long sequences of tokens from a large vocabulary, making it difficult to integrate it at low cost in a large language model. By applying a simple lasso-based feature selection on multimodal Bag-of-Words representation, we retain only the most important audio tokens for the task, and adapt the language model to them with a self-supervised language modeling objective, before fine-tuning it on the downstream task. We show this helps to improve the performances compared to an unimodal model, to a bigger SpeechLM or to integrating audio via a learned representation. We show the effectiveness of our method on two recent Argumentative Fallacy Detection and Classification tasks where the use of audio was believed counterproductive, reaching state-of-the-art results. We also provide an in-depth analysis of the method, showing that even a random audio token selection helps enhancing the unimodal model. Our code is available [online](https://github.com/salocinc/EACL26SpeechTokFallacy/).

A multimodal Bayesian Network for symptom-level depression and anxiety prediction from voice and speech data

During psychiatric assessment, clinicians observe not only what patients report, but important nonverbal signs such as tone, speech rate, fluency, responsiveness, and body language. Weighing and integrating these different information sources is a challenging task and a good candidate for support by intelligence-driven tools - however this is yet to be realized in the clinic. Here, we argue that several important barriers to adoption can be addressed using Bayesian network modelling. To demonstrate this, we evaluate a model for depression and anxiety symptom prediction from voice and speech features in large-scale datasets (30,135 unique speakers). Alongside performance for conditions and symptoms (for depression, anxiety ROC-AUC=0.842,0.831 ECE=0.018,0.015; core individual symptom ROC-AUC>0.74), we assess demographic fairness and investigate integration across and redundancy between different input modality types. Clinical usefulness metrics and acceptability to mental health service users are explored. When provided with sufficiently rich and large-scale multimodal data streams and specified to represent common mental conditions at the symptom rather than disorder level, such models are a principled approach for building robust assessment support tools: providing clinically-relevant outputs in a transparent and explainable format that is directly amenable to expert clinical supervision.

Scriboora: Rethinking Human Pose Forecasting

Human pose forecasting predicts future poses based on past observations, and has many significant applications in areas such as action recognition, autonomous driving or human-robot interaction. This paper evaluates a wide range of pose forecasting algorithms in the task of absolute pose forecasting, revealing many reproducibility issues, and provides a unified training and evaluation pipeline. After drawing a high-level analogy to the task of speech understanding, it is shown that recent speech models can be efficiently adapted to the task of pose forecasting, and improve current state-of-the-art performance. At last the robustness of the models is evaluated, using noisy joint coordinates obtained from a pose estimator model, to reflect a realistic type of noise, which is more close to real-world applications. For this a new dataset variation is introduced, and it is shown that estimated poses result in a substantial performance degradation, and how much of it can be recovered again by unsupervised finetuning.

PresentCoach: Dual-Agent Presentation Coaching through Exemplars and Interactive Feedback

Effective presentation skills are essential in education, professional communication, and public speaking, yet learners often lack access to high-quality exemplars or personalized coaching. Existing AI tools typically provide isolated functionalities such as speech scoring or script generation without integrating reference modeling and interactive feedback into a cohesive learning experience. We introduce a dual-agent system that supports presentation practice through two complementary roles: the Ideal Presentation Agent and the Coach Agent. The Ideal Presentation Agent converts user-provided slides into model presentation videos by combining slide processing, visual-language analysis, narration script generation, personalized voice synthesis, and synchronized video assembly. The Coach Agent then evaluates user-recorded presentations against these exemplars, conducting multimodal speech analysis and delivering structured feedback in an Observation-Impact-Suggestion (OIS) format. To enhance the authenticity of the learning experience, the Coach Agent incorporates an Audience Agent, which simulates the perspective of a human listener and provides humanized feedback reflecting audience reactions and engagement. Together, these agents form a closed loop of observation, practice, and feedback. Implemented on a robust backend with multi-model integration, voice cloning, and error handling mechanisms, the system demonstrates how AI-driven agents can provide engaging, human-centered, and scalable support for presentation skill development in both educational and professional contexts.

Auden-Voice: General-Purpose Voice Encoder for Speech and Language Understanding

Human voice encodes both identity and paralinguistic cues, yet encoders in large audio-language models (LALMs) rarely balance both aspects. In this work, we present a study toward building a general-purpose voice encoder that captures nuanced voice cues. Through a comprehensive evaluation, we find that multi-task training yields the most balanced representations, whereas contrastive language-audio pretraining (CLAP) primarily improves retrieval without enhancing paralinguistic understanding. Our final encoder, Auden-Voice, also demonstrates strong performance when integrated with LLMs. The code and training recipes will be released with the audio understanding toolkit Auden.