arXiv subjects

Samuele Cornell

Publications and source records attributed to Samuele Cornell.

At least 19 recordsLinked to original sources

arXiv2026

Dissecting Sensitivity to Training Language in Self-Supervised Speech Learning Using Neural Audio Codec Tokens

Neural audio codecs (NACs) have become popular for obtaining speech representations as discrete tokens. Beyond compression, discrete tokens can be used to train self-supervised learning (SSL) models. Such models, referred to as codec-based SSL models, reduce data storage and computational cost, enabling scalable SSL pre-training. However, their language sensitivity remains unclear. When the language changes, codec-based SSL models may require retraining, which undermines their efficiency. In this paper, we present a systematic analysis of language sensitivity by varying either the NAC training language or the SSL pre-training language while keeping the other fixed. Experimental results show that downstream performance is insensitive to the NAC training language but strongly dependent on the SSL pre-training language. These findings suggest that a single NAC can be reused across languages, while aligning the SSL pre-training language with the target language is crucial.

cs.SD↗

arXiv2026

ESPnet3: Infrastructure for Scalable Speech and Audio Research in the Foundation Model Era

Recent speech research involves increasingly large datasets, complex models, and diverse experimental workflows. However, existing frameworks require substantial engineering effort to support such experiments. We present ESPnet3, a speech and audio research framework built on a modular system architecture with configuration-driven dataset composition and unified Python-based workflows. ESPnet3 introduces a DataOrganizer abstraction for flexible dataset integration and dataset sharding for memory-efficient large-scale training, while allowing recipe-specific logic through lightweight stage overrides. In OWSM pre-training experiments, ESPnet3 reduces per-epoch training time by \emph{21.1 minutes} compared to ESPnet2 and achieves \emph{>80\% GPU utilization} in multi-node training. Fine-tuning experiments show that new models and datasets can be integrated with around \emph{46 lines of additional code}. ESPnet3 will be publicly released with model checkpoints and training logs.

eess.AS↗

arXiv2026

Grounding Spoken LLMs in Multi-Speaker Audio via Diarization Conditioning

We propose diarization-conditioned spoken language models (SLMs), a strategy for extending SLMs to far-field multi-talker audio. Rather than adapting the decoder via Serialized Output Training, which risks catastrophic forgetting, we condition the acoustic encoder on diarization masks to extract target-speaker representations, keeping the decoder frozen. We instantiate this as Dixtral, integrating a Diarization Conditioned Whisper (DiCoW) encoder into the Voxtral SLM. On AMI, NOTSOFAR-1, LibriSpeechMix, and Mixer6, Dixtral outperforms Gemini 3.0 Flash, VibeVoice, and Voxtral Mini Transcribe V2 on speaker-attributed transcription by 29.0%, 19.8%, and 16.0% absolute cpWER respectively. On a novel long-form multi-speaker QA benchmark, zero-shot Dixtral matches Gemini on far-field content understanding, and when fine-tuned surpasses both Gemini and Voxtral operating on close-talk across all tasks.

eess.AS↗

arXiv2026

Exploiting Noise Inseparability for Weakly-Supervised Discriminative Speech Denoising Using Noisy Targets

Speech denoising is an often necessary step not only for human listening, but also for downstream processing by systems lacking robustness to noisy, real-world acoustic conditions. Unfortunately, denoising is a problem where conventional in-domain supervised training is not trivial, as the training targets cannot be annotated by humans: producing a clean version of a naturally-noisy speech recording is itself the task to solve. Supervised training is typically performed through the artificial addition of noise to clean speech recordings, which can only be sourced from controlled domains, a significant limitation due to the poor out-of-domain generalization of neural networks. An alternative is noisy target training (NyTT), which simply replaces the clean speech with in-domain noisy recordings, with the hope that learning to remove the artificial noise will extend to the natural. Though having shown promising results, NyTT's training objective is not minimized by clean speech estimates. We show that by estimating the artificial noise in addition to the naturally-noisy speech, the undesirable optimum can actually be exploited: the residual noise in the speech estimate can be canceled by the noise estimate via simple subtraction. Crucially, the optimum is fully compatible with conventional artificial mixtures, enabling joint training using both types of data with consistent optimization targets, opening the door to improved domain adaptability. The effectiveness of our approach is demonstrated through WHAM! and CHiME-3-based benchmarks.

eess.AS↗

arXiv2026

Cross-Talk Speech Reduction, by Separation, for Separation

In conversational speech separation and recognition tasks, close-talk microphones are typically attached to each speaker during training data collection to capture near-field, close-talk mixture signals, in addition to using far-field microphones to record far-field mixture signals. Each such close-talk mixture exhibits a reasonably high energy level for the wearer and could intuitively serve as weak supervision for training far-field speech separation models directly on real-recorded far-field signals. However, they are not sufficiently clean for this purpose, as they often contain strong cross-talk speech from other speakers in addition to background noise. To address this, we propose cross-talk reduction (CTR), a task aiming to isolate the wearer's speech from each close-talk mixture, and a novel method called CTRnet, which can be trained directly on real-recorded pairs of close-talk and far-field mixtures to accomplish CTR. Building on CTRnet, we further propose pseudo-label based far-field speech separation (PuLSS), which uses CTRnet's estimated clean speech as pseudo-labels to train models for separating far-field mixtures. A key advantage of the proposed framework is that both CTRnet and PuLSS can be trained on real-recorded data from the target domain, addressing the generalization gap commonly observed when models are trained exclusively on simulated data. On the CHiME-6 dataset, our framework achieves state-of-the-art ASR performance under both oracle and estimated speaker diarization, surpassing all CHiME-{7,8} challenge submissions. To our knowledge, it is the first neural speech separation method that substantially outperforms guided source separation on real conversational "speech-in-the-wild" data.

eess.AS↗

arXiv2026

PlanRAG-Audio: Planning and Retrieval Augmented Generation for Long-form Audio Understanding

Long-form audio understanding poses significant challenges for large audio language models (LALMs) due to the extreme length of audio sequences and the need to reason over heterogeneous acoustic cues distributed over time, such as speech content, speaker identity, emotion, and sound events. To address these challenges, we propose \textbf{PlanRAG-Audio}, a planning-based retrieval-augmented generation framework for scalable long-form audio understanding. Rather than having audio LALMs process entire recordings directly, PlanRAG-Audio explicitly plans which modalities and temporal spans are required for a given query, and retrieves only query-relevant information from a structured text and audio database. This retrieval planning enables effective reasoning over complex, cross-domain audio queries while substantially reducing the input length passed to the large language models. Experiments across a wide range of speech/audio retrieval demonstrate that PlanRAG-Audio improves reasoning accuracy and stabilizes performance as audio duration increases by decoupling inference cost from raw audio length.

eess.AS↗

arXiv2026

Mind the Gap: Impact of Synthetic Conversational Data on Multi-Talker ASR and Speaker Diarization

Recent breakthroughs in multi-talker ASR (MT-ASR) and speaker diarization (SD) rely on synthetic data to mitigate the scarcity of large-scale conversational recordings, yet the impact of specific simulation choices remains poorly understood. To mind the gap between simulated mixtures and real-world interactions, we present a study of synthetic data generation for leading MT-ASR (DiCoW) and SD (Sortformer) systems. By introducing FastMSS, a highly efficient open-source simulator, we analyze turn-taking dynamics, source domain, acoustic augmentation, and data mixing strategies. Our findings reveal that optimal simulation recipes are highly task-dependent: increasing speech overlap benefits ASR but degrades diarization. Furthermore, broad source diversity consistently outperforms exact domain matching. Ultimately, synthetic-only training approaches real-data baselines, and combining simulated data with real recordings yields substantial gains over real-only training across both tasks.

eess.AS↗

arXiv2026

Ring Mixing with Auxiliary Signal-to-Consistency-Error Ratio Loss for Unsupervised Denoising in Speech Separation

Noisy speech separation systems are typically trained on fully-synthetic mixtures, limiting generalization to real-world scenarios. Though training on mixtures of in-domain (thus often noisy) speech is possible, we show that this leads to undesirable optima where mixture noise is retained in the estimates, due to the inseparability of the background noises and the loss function's symmetry. To address this, we propose ring mixing, a batch strategy of using each source in two mixtures, alongside a new Signal-to-Consistency-Error Ratio (SCER) auxiliary loss penalizing inconsistent estimates of the same source from different mixtures, breaking symmetry and incentivizing denoising. On a WHAM!-based benchmark, our method can reduce residual noise by upwards of half, effectively learning to denoise from only noisy recordings. This opens the door to training more generalizable systems using in-the-wild data, which we demonstrate via systems trained using naturally-noisy speech from VoxCeleb.

eess.AS↗

arXiv2026

Who Spoke What When? Evaluating Spoken Language Models for Conversational ASR with Semantic and Overlap-Aware Metrics

Conversational automatic speech recognition remains challenging due to overlapping speech, far-field noise, and varying speaker counts. While recent LLM-based systems perform well on single-speaker benchmarks, their robustness in multi-speaker settings is unclear. We systematically compare LLM-based and modular pipeline approaches along four axes: overlap robustness, semantic fidelity, speaker count, and single- versus multi-channel input. To capture meaning-altering errors that conventional metrics miss, we introduce tcpSemER, which extends tcpWER by replacing Levenshtein distance with embedding-based semantic similarity. We further decompose tcpWER into overlapping and non-overlapping components for finer-grained analysis. Experiments across three datasets show that LLM-based systems are competitive in two-speaker settings but degrade as speaker count and overlap increase, whereas modular pipelines remain more robust.

cs.CL↗

arXiv2026

Modeling Overlapped Speech with Shuffles

We propose to model parallel streams of data, such as overlapped speech, using shuffles. Specifically, this paper shows how the shuffle product and partial order finite-state automata (FSAs) can be used for alignment and speaker-attributed transcription of overlapped speech. We train using the total score on these FSAs as a loss function, marginalizing over all possible serializations of overlapping sequences at subword, word, and phrase levels. To reduce graph size, we impose temporal constraints by constructing partial order FSAs. We address speaker attribution by modeling (token, speaker) tuples directly. Viterbi alignment through the shuffle product FSA directly enables one-pass alignment. We evaluate performance on synthetic LibriSpeech overlaps. To our knowledge, this is the first algorithm that enables single-pass alignment of multi-talker recordings. All algorithms are implemented using k2 / Icefall.

cs.SD↗

arXiv2026

SE-DiCoW: Self-Enrolled Diarization-Conditioned Whisper

Speaker-attributed automatic speech recognition (ASR) in multi-speaker environments remains a major challenge. While some approaches achieve strong performance when fine-tuned on specific domains, few systems generalize well across out-of-domain datasets. Our prior work, Diarization-Conditioned Whisper (DiCoW), leverages speaker diarization outputs as conditioning information and, with minimal fine-tuning, demonstrated strong multilingual and multi-domain performance. In this paper, we address a key limitation of DiCoW: ambiguity in Silence-Target-Non-target-Overlap (STNO) masks, where two or more fully overlapping speakers may have nearly identical conditioning despite differing transcriptions. We introduce SE-DiCoW (Self-Enrolled Diarization-Conditioned Whisper), which uses diarization output to locate an enrollment segment anywhere in the conversation where the target speaker is most active. This enrollment segment is used as fixed conditioning via cross-attention at each encoder layer. We further refine DiCoW with improved data segmentation, model initialization, and augmentation. Together, these advances yield substantial gains: SE-DiCoW reduces macro-averaged tcpWER by 52.4% relative to the original DiCoW on the EMMA MT-ASR benchmark.

eess.AS↗

arXiv2026

UrgentMOS: Unified Multi-Metric and Preference Learning for Robust Speech Quality Assessment

Automatic speech quality assessment has become increasingly important as modern speech generation systems continue to advance, while human listening tests remain costly, time-consuming, and difficult to scale. Most existing learning-based assessment models rely primarily on scarce human-annotated mean opinion score (MOS) data, which limits robustness and generalization, especially when training across heterogeneous datasets. In this work, we propose UrgentMOS, a unified speech quality assessment framework that jointly learns from diverse objective and perceptual quality metrics, while explicitly tolerating the absence of arbitrary subsets of metrics during training. By leveraging complementary quality facets under heterogeneous supervision, UrgentMOS enables effective utilization of partially annotated data and improves robustness when trained on large-scale, multi-source datasets. Beyond absolute score prediction, UrgentMOS explicitly models pairwise quality preferences by directly predicting comparative MOS (CMOS), making it well suited for preference-based evaluation scenarios commonly adopted in system benchmarking. Extensive experiments across a wide range of speech quality datasets, including simulated distortions, speech enhancement, and speech synthesis, demonstrate that UrgentMOS consistently achieves state-of-the-art performance in both absolute and comparative evaluation settings.

cs.SD↗

arXiv2026

The CMU-AIST submission for the ICME 2025 Audio Encoder Challenge

This technical report describes our submission to the ICME 2025 audio encoder challenge. Our submitted system is built on BEATs, a masked speech token prediction based audio encoder. We extend the BEATs model using 74,000 hours of data derived from various speech, music, and sound corpora and scale its architecture upto 300 million parameters. We experiment with speech-heavy and balanced pre-training mixtures to study the impact of different domains on final performance. Our submitted system consists of an ensemble of the Dasheng 1.2 billion model with two custom scaled-up BEATs models trained on the aforementioned pre-training data mixtures. We also propose a simple ensembling technique that retains the best capabilities of constituent models and surpasses both the baseline and Dasheng 1.2B. For open science, we publicly release our trained checkpoints via huggingface at https://huggingface.co/shikhar7ssu/OpenBEATs-ICME-SOUND and https://huggingface.co/shikhar7ssu/OpenBEATs-ICME.

cs.SD↗

arXiv2026

ICASSP 2026 URGENT Speech Enhancement Challenge

The ICASSP 2026 URGENT Challenge advances the series by focusing on universal speech enhancement (SE) systems that handle diverse distortions, domains, and input conditions. This overview paper details the challenge's motivation, task definitions, datasets, baseline systems, evaluation protocols, and results. The challenge is divided into two complementary tracks. Track 1 focuses on universal speech enhancement, while Track 2 introduces speech quality assessment for enhanced speech. The challenge attracted over 80 team registrations, with 29 submitting valid entries, demonstrating significant community interest in robust SE technologies.

eess.AS↗

arXiv2025

CS-FLEURS: A Massively Multilingual and Code-Switched Speech Dataset

We present CS-FLEURS, a new dataset for developing and evaluating code-switched speech recognition and translation systems beyond high-resourced languages. CS-FLEURS consists of 4 test sets which cover in total 113 unique code-switched language pairs across 52 languages: 1) a 14 X-English language pair set with real voices reading synthetically generated code-switched sentences, 2) a 16 X-English language pair set with generative text-to-speech 3) a 60 {Arabic, Mandarin, Hindi, Spanish}-X language pair set with the generative text-to-speech, and 4) a 45 X-English lower-resourced language pair test set with concatenative text-to-speech. Besides the four test sets, CS-FLEURS also provides a training set with 128 hours of generative text-to-speech data across 16 X-English language pairs. Our hope is that CS-FLEURS helps to broaden the scope of future code-switched speech research. Dataset link: https://huggingface.co/datasets/byan/cs-fleurs.

cs.CL↗

arXiv2025

MAPSS: Manifold-based Assessment of Perceptual Source Separation

Objective assessment of audio source-separation systems still mismatches subjective human perception, especially when interference from competing talkers and distortion of the target signal interact. We introduce Perceptual Separation (PS) and Perceptual Match (PM), a complementary pair of measures that, by design, isolate these leakage and distortion factors. Our intrusive approach generates a set of fundamental distortions, e.g., clipping, notch filter, and pitch shift from each reference waveform signal in the mixture. Distortions, references, and system outputs from all sources are independently encoded by a pre-trained self-supervised model, then aggregated and embedded with a manifold learning technique called diffusion maps, which aligns Euclidean distances on the manifold with dissimilarities of the encoded waveform representations. On this manifold, PM captures the self-distortion of a source by measuring distances from its output to its reference and associated distortions, while PS captures leakage by also accounting for distances from the output to non-attributed references and distortions. Both measures are differentiable and operate at a resolution as high as 75 frames per second, allowing granular optimization and analysis. We further derive, for both measures, frame-level deterministic error radius and non-asymptotic, high-probability confidence intervals. Experiments on English, Spanish, and music mixtures show that, against 18 widely used measures, the PS and PM are almost always placed first or second in linear and rank correlations with subjective human mean-opinion scores.

eess.AS↗

arXiv2025

Recent Trends in Distant Conversational Speech Recognition: A Review of CHiME-7 and 8 DASR Challenges

The CHiME-7 and 8 distant speech recognition (DASR) challenges focus on multi-channel, generalizable, joint automatic speech recognition (ASR) and diarization of conversational speech. With participation from 9 teams submitting 32 diverse systems, these challenges have contributed to state-of-the-art research in the field. This paper outlines the challenges' design, evaluation metrics, datasets, and baseline systems while analyzing key trends from participant submissions. From this analysis it emerges that: 1) Most participants use end-to-end (e2e) ASR systems, whereas hybrid systems were prevalent in previous CHiME challenges. This transition is mainly due to the availability of robust large-scale pre-trained models, which lowers the data burden for e2e-ASR. 2) Despite recent advances in neural speech separation and enhancement (SSE), all teams still heavily rely on guided source separation, suggesting that current neural SSE techniques are still unable to reliably deal with complex scenarios and different recording setups. 3) All best systems employ diarization refinement via target-speaker diarization techniques. Accurate speaker counting in the first diarization pass is thus crucial to avoid compounding errors and CHiME-8 DASR participants especially focused on this part. 4) Downstream evaluation via meeting summarization can correlate weakly with transcription quality due to the remarkable effectiveness of large-language models in handling errors. On the NOTSOFAR-1 scenario, even systems with over 50% time-constrained minimum permutation WER can perform roughly on par with the most effective ones (around 11%). 5) Despite recent progress, accurately transcribing spontaneous speech in challenging acoustic environments remains difficult, even when using computationally intensive system ensembles.

eess.AS↗

arXiv2025

OpenBEATs: A Fully Open-Source General-Purpose Audio Encoder

Masked token prediction has emerged as a powerful pre-training objective across language, vision, and speech, offering the potential to unify these diverse modalities through a single pre-training task. However, its application for general audio understanding remains underexplored, with BEATs being the only notable example. BEATs has seen limited modifications due to the absence of open-source pre-training code. Furthermore, BEATs was trained only on AudioSet, restricting its broader downstream applicability. To address these gaps, we present OpenBEATs, an open-source framework that extends BEATs via multi-domain audio pre-training. We conduct comprehensive evaluations across six types of tasks, twenty five datasets, and three audio domains, including audio reasoning tasks such as audio question answering, entailment, and captioning. OpenBEATs achieves state-of-the-art performance on six bioacoustics datasets, two environmental sound datasets and five reasoning datasets, performing better than models exceeding a billion parameters at one-fourth their parameter size. These results demonstrate the effectiveness of multi-domain datasets and masked token prediction task to learn general-purpose audio representations. To promote further research and reproducibility, we release all pre-training and evaluation code, pretrained and fine-tuned checkpoints, and training logs at https://github.com/Shikhar-S/OpenBEATs

cs.SD↗