arXiv subjects

Joseph Keshet

Publications and source records attributed to Joseph Keshet.

At least 19 recordsLinked to original sources

arXiv2026

Hidden in the Request: Explaining Unethical LLM Compliance through Token Relevance

Although Large Language Models (LLMs) are aligned to optimize for both helpfulness and harmlessness, these dual objectives may conflict, inevitably leading to alignment failures. This work systematically investigates instances where LLMs fail to exhibit ethical behavior. To understand the underlying mechanics of these vulnerabilities, we introduce a probing methodology that presents unethical scenarios to LLMs in three distinct structural modalities: objective classification tasks, subjective first-person statements, and direct requests for assistance. We find that model performance degrades in the request-for-assistance-based form. Using Layer-wise Relevance Propagation (LRP), we trace this discrepancy to an attribution bias: the model places greater emphasis on benign task-framing tokens (e.g., "Can you help me...") than on tokens signaling the underlying unethical behavior (e.g., "without getting caught"), which we term cue-tokens. We hypothesize that this under-attribution contributes to harmful compliance. To test this, we introduce two LRP-guided decoding methods that steer generation toward trajectories more relevant to cue tokens. Empirical evaluations show that these interventions promote safer responses, supporting cue-token attribution's role in compliance failures.

cs.AI↗

arXiv2026

Self-Supervised Test-Time Tuning for Packet Loss Concealment

Packet loss concealment (PLC) reconstructs audio packets that are missing at the receiver, usually with a trained model whose parameters remain fixed at deployment time. This treats the PLC model as static, even though each call or recording exposes signal-specific information through the packets that did arrive. We present TTT-PLC, a self-supervised test-time tuning framework that adapts existing PLC models using only those received packets. The method creates supervision by synthetically masking portions of the available signal, training the model to conceal them with its native PLC objective, and then using the adapted model to reconstruct the true packet losses. No clean reference signal, external adaptation data, or architectural modification is required. We study TTT-PLC in two deployment settings. In the non-causal setting, the received file is available before reconstruction, allowing repeated self-supervised adaptation passes and providing a per-file adaptation ceiling. In the causal setting, audio is streamed without revising emitted samples; adaptation is performed only on completed past blocks, and updated parameters affect only future audio. We instantiate the framework on two public PLC backbones, FRN, a recurrent full-band speech PLC model, and PARCnet, a hybrid autoregressive-neural model for networked music. Across these settings, the results show that pretrained PLC systems do not need to be treated as fixed at inference time, the still-observed portions of a lossy signal can provide an effective training signal for improving concealment on that same signal.

eess.AS↗

arXiv2026

Fully Differentiable Neural Forced Alignment via Soft Dynamic Programming

Recent advances in sequence modeling have significantly improved ASR systems, bringing them close to human-level recognition accuracy and enhancing robustness across diverse acoustic conditions and languages. In contrast, Forced Alignment has not experienced comparable progress, and traditional HMM-GMM frameworks remain widely adopted and highly competitive. To address this gap, we propose an end-to-end, fully differentiable neural architecture specifically designed for phoneme alignment. The model consists of an encoder that processes the input signal and a decoder that produces alignment decisions. The encoder is structured into two complementary branches: one dedicated to phoneme identity verification and the other to phoneme boundary detection. The decoder is implemented as a trainable module based on differentiable soft dynamic programming. The entire system is optimized end-to-end using a novel contrastive loss that encourages clear separation between steady-state phoneme regions and transition boundaries. The proposed approach outperforms the current state of the art in phoneme alignment on hand-annotated English benchmarks, achieves strong word-level generalization results, and demonstrates generalization on unseen languages.

eess.AS↗

arXiv2026

Why Sampling Is Not Choosing: Intentionality, Agency, and Moral Responsibility in Large Language Models

Recent advances in large language models (LLMs) have prompted claims that such systems exhibit agency or qualify as moral agents. This paper argues that these attributions are misguided. We maintain that moral responsibility requires commitment-bearing agency grounded in intrinsic intentionality and self-attributed action, and that such agency constitutes the form of free will relevant to responsibility. Although LLMs generate coherent and normatively evaluable outputs, their operation is fully characterized by probabilistic input-output mappings learned from data. Their apparent intentionality is derived rather than intrinsic, and their outputs are neither owned as commitments nor guided by reasons. Variability introduced by stochastic sampling does not amount to choice or authorship. We address objections from the intentional stance, functionalism, compatibilism, and the presence of moral reasoning in model outputs, arguing that none suffice to establish genuine agency.

cs.AI↗

arXiv2026

Multilingual Word-Level Forced Alignment with Self-Supervised Representations and Learned Dynamic Programming

We present a method for accurate multilingual word-level forced alignment, consisting of an alignment encoder and a learned alignment decoder. The encoder integrates two representations: one from the Massively Multilingual Speech (MMS) model and another from a self-supervised phoneme boundary detector (UnSupSeg). It learns to fuse them and to estimate word-boundary probabilities over long temporal contexts. The alignment decoder is a learned dynamic programming that combines encoder outputs with segmental features over the MMS and UnSupSeg representations to infer final word boundaries. Trained iteratively on TIMIT and Buckeye, the proposed approach outperforms Montreal Forced Aligner (MFA) and MMS-based alignment on both datasets. On unseen languages (Dutch, German, and Hebrew), the proposed model achieves performance consistently better than or on par with existing alignment approaches, indicating its potential to scale to 1100+ languages supported by MMS without further training.

cs.CL↗

arXiv2026

Mitigating Multimodal LLMs Hallucinations via Relevance Propagation at Inference Time

Multimodal large language models (MLLMs) have revolutionized the landscape of AI, demonstrating impressive capabilities in tackling complex vision and audio-language tasks. However, a critical challenge remains: these models often suffer from hallucinations, generating outputs that diverge from the provided perceptual inputs. This tendency stems from an inherent imbalance in modality utilization during inference, where the dominance of textual tokens undermines the potential of perceptual inputs. As a result, the model frequently resorts to textual language priors at the expense of grounded evidence. To tackle this issue, we propose Learning Inference-time Modality Enhancement (LIME), a training-free framework designed to bolster multimodal grounding by explicitly enhancing modality usage during decoding. LIME leverages Layer-wise Relevance Propagation (LRP) to quantify token-level contributions and defines a relevance-based objective that promotes increased reliance on perceptual inputs. This objective is enforced through inference-time updates to the model's key-value representations, without modifying model parameters or requiring additional training data. We evaluate LIME across multiple multimodal benchmarks in both vision and audio domains, demonstrating consistent reductions in hallucinations and enhanced grounding while preserving generation quality. Further analysis shows that LIME increases modality contribution and produces more localized and semantically aligned relevance patterns.

cs.LG↗

arXiv2026

Joint Enhancement and Classification using Coupled Diffusion Models of Signals and Logits

Robust classification in noisy environments remains a fundamental challenge in machine learning. Standard approaches typically treat signal enhancement and classification as separate, sequential stages: first enhancing the signal and then applying a classifier. This approach fails to leverage the semantic information in the classifier's output during denoising. In this work, we propose a general, domain-agnostic framework that integrates two interacting diffusion models: one operating on the input signal and the other on the classifier's output logits, without requiring any retraining or fine-tuning of the classifier. This coupled formulation enables mutual guidance, where the enhancing signal refines the class estimation and, conversely, the evolving class logits guide the signal reconstruction towards discriminative regions of the manifold. We introduce three strategies to effectively model the joint distribution of the input and the logit. We evaluated our joint enhancement method for image classification and automatic speech recognition. The proposed framework surpasses traditional sequential enhancement baselines, delivering robust and flexible improvements in classification accuracy under diverse noise conditions.

cs.LG↗

arXiv2026

Analyzing and Guiding Zero-Shot Posterior Sampling in Diffusion Models

Recovering a signal from its degraded measurements is a long standing challenge in science and engineering. Recently, zero-shot diffusion based methods have been proposed for such inverse problems, offering a posterior sampling based solution that leverages prior knowledge. Such algorithms incorporate the observations through inference, often leaning on manual tuning and heuristics. In this work we propose a rigorous analysis of these approximate posterior samplers, relying on a Gaussianity assumption of the prior. Under this regime, we show that both the ideal posterior sampler and diffusion-based reconstruction algorithms can be expressed in closed-form, enabling their thorough analysis and comparisons in the spectral domain. Building on these representations, we introduce a principled framework for parameter design, replacing heuristic selection strategies used to date. The proposed approach is method-agnostic and yields tailored parameter choices that jointly account for the characteristics of the prior, the degraded signal, and the diffusion dynamics. We show that our spectral recommendations differ structurally from standard heuristics and vary with the diffusion step size, resulting in a consistent balance between perceptual quality and signal fidelity.

cs.LG↗

arXiv2025

Drax: Speech Recognition with Discrete Flow Matching

Diffusion and flow-based non-autoregressive (NAR) models have shown strong promise in large language modeling, however, their potential for automatic speech recognition (ASR) remains largely unexplored. We propose Drax, a discrete flow matching framework for ASR that enables efficient parallel decoding. To better align training with inference, we construct an audio-conditioned probability path that guides the model through trajectories resembling likely intermediate inference errors, rather than direct random noise to target transitions. Our theoretical analysis links the generalization gap to divergences between training and inference occupancies, controlled by cumulative velocity errors, thereby motivating our design choice. Empirical evaluation demonstrates that our approach attains recognition accuracy on par with state-of-the-art speech models while offering improved accuracy-efficiency trade-offs, highlighting discrete flow matching as a promising direction for advancing NAR ASR.

eess.AS↗

arXiv2025

Beyond Transcription: Mechanistic Interpretability in ASR

Interpretability methods have recently gained significant attention, particularly in the context of large language models, enabling insights into linguistic representations, error detection, and model behaviors such as hallucinations and repetitions. However, these techniques remain underexplored in automatic speech recognition (ASR), despite their potential to advance both the performance and interpretability of ASR systems. In this work, we adapt and systematically apply established interpretability methods such as logit lens, linear probing, and activation patching, to examine how acoustic and semantic information evolves across layers in ASR systems. Our experiments reveal previously unknown internal dynamics, including specific encoder-decoder interactions responsible for repetition hallucinations and semantic biases encoded deep within acoustic representations. These insights demonstrate the benefits of extending and applying interpretability techniques to speech recognition, opening promising directions for future research on improving model transparency and robustness.

cs.SD↗

arXiv2025

WhisperRT -- Turning Whisper into a Causal Streaming Model

Automatic Speech Recognition (ASR) has seen remarkable progress, with models like OpenAI Whisper and NVIDIA Canary achieving state-of-the-art (SOTA) performance in offline transcription. However, these models are not designed for streaming (online or real-time) transcription, due to limitations in their architecture and training methodology. We propose a method to turn the transformer encoder-decoder model into a low-latency streaming model. The encoder is made causal to process audio incrementally, while the decoder conditions on partial encoder states to generate tokens aligned with the available temporal context. This requires explicit synchronization between encoded input frames and token emissions. Since tokens are produced only after sufficient acoustic evidence is observed, an inherent latency arises, necessitating fine-tuning of the encoder-decoder alignment mechanism. We propose an updated inference mechanism that utilizes the fine-tuned causal encoder and decoder to yield greedy and beam-search decoding, and is shown to be locally optimal. Experiments on low-latency chunk sizes (less than 300 msec) show that our fine-tuned model outperforms existing non-fine-tuned streaming approaches in most cases, while using a lower complexity. We release our training and inference code, along with the fine-tuned models, to support further research and development in streaming ASR.

cs.CL↗

arXiv2025

How Does a Deep Neural Network Look at Lexical Stress in English Words?

Despite their success in speech processing, neural networks often operate as black boxes, prompting the question: what informs their decisions, and how can we interpret them? This work examines this issue in the context of lexical stress. A dataset of English disyllabic words was automatically constructed from read and spontaneous speech. Several Convolutional Neural Network (CNN) architectures were trained to predict stress position from a spectrographic representation of disyllabic words lacking minimal stress pairs (e.g., initial stress WAllet, final stress exTEND), achieving up to 92% accuracy on held-out test data. Layerwise Relevance Propagation (LRP), a technique for neural network interpretability analysis, revealed that predictions for held-out minimal pairs (PROtest vs. proTEST ) were most strongly influenced by information in stressed versus unstressed syllables, particularly the spectral properties of stressed vowels. However, the classifiers also attended to information throughout the word. A feature-specific relevance analysis is proposed, and its results suggest that our best-performing classifier is strongly influenced by the stressed vowel's first and second formants, with some evidence that its pitch and third formant also contribute. These results reveal deep learning's ability to acquire distributed cues to stress from naturally occurring data, extending traditional phonetic work based around highly controlled stimuli.

cs.CL↗

arXiv2025

Keyword Spotting with Hyper-Matched Filters for Small Footprint Devices

Open-vocabulary keyword spotting (KWS) refers to the task of detecting words or terms within speech recordings, regardless of whether they were included in the training data. This paper introduces an open-vocabulary keyword spotting model with state-of-the-art detection accuracy for small-footprint devices. The model is composed of a speech encoder, a target keyword encoder, and a detection network. The speech encoder is either a tiny Whisper or a tiny Conformer. The target keyword encoder is implemented as a hyper-network that takes the desired keyword as a character string and generates a unique set of weights for a convolutional layer, which can be considered as a keyword-specific matched filter. The detection network uses the matched-filter weights to perform a keyword-specific convolution, which guides the cross-attention mechanism of a Perceiver module in determining whether the target term appears in the recording. The results indicate that our system achieves state-of-the-art detection performance and generalizes effectively to out-of-domain conditions, including second-language (L2) speech. Notably, our smallest model, with just 4.2 million parameters, matches or outperforms models that are several times larger, demonstrating both efficiency and robustness.

eess.AS↗

arXiv2025

PatchDSU: Uncertainty Modeling for Out of Distribution Generalization in Keyword Spotting

Deep learning models excel at many tasks but rely on the assumption that training and test data follow the same distribution. This assumption often does not hold in real-world speech systems, where distribution shifts are common due to varying environments, recording conditions, and speaker diversity. The method of Domain Shifts with Uncertainty (DSU) augments the input of each neural network layer based on the input feature statistics. It addresses the problem of out-of-domain generalization by assuming feature statistics follow a multivariate Gaussian distribution and substitutes the input with sampled features from this distribution. While effective for computer vision, applying DSU to speech presents challenges due to the nature of the data. Unlike static visual data, speech is a temporal signal commonly represented by a spectrogram - the change of frequency over time. This representation cannot be treated as a simple image, and the resulting sparsity can lead to skewed feature statistics when applied to the entire input. To tackle out-of-distribution issues in keyword spotting, we propose PatchDSU, which extends DSU by splitting the input into patches and independently augmenting each patch. We evaluated PatchDSU and DSU alongside other methods on the Google Speech Commands, Librispeech, and TED-LIUM. Additionally, we evaluated performance under white Gaussian and MUSAN music noise conditions. We also explored out-of-domain generalization by analyzing model performance on datasets they were not trained on. Overall, in most cases, both PatchDSU and DSU outperform other methods. Notably, PatchDSU demonstrates more consistent improvements across the evaluated scenarios compared to other approaches.

eess.AS↗

arXiv2025

Large Language Model Guided Decoding for Self-Supervised Speech Recognition

Self-supervised automatic speech recognition (SSL-ASR) is an ASR approach that uses speech encoders pretrained on large amounts of unlabeled audio (e.g., wav2vec2.0 or HuBERT) and then fine-tunes them with limited labeled data to perform transcription. Decoding is usually performed with a CTC decoder, whose hypotheses are scored and refined using an external language model (LM), typically an n-gram or neural LM, which guides beam search to produce the final transcription. Using Large Language Models (LLMs) as external LMs remains a challenge, as their word probabilities are overly confident. The proposed method integrates an LLM with an SSL acoustic model by using the LLM's decoding mechanism to generate a set of candidate next tokens. For each candidate, the SSL model provides an acoustic score by aligning it to the input acoustics of the SSL model. A combined acoustic and LLM score is then calculated based on decomposing the MAP estimator of words given the acoustic signal. The tokens with the highest combined scores are maintained in a beam, which is then used to proceed to the next decoding step. We illustrate the effectiveness of our method through a comprehensive comparison with the current state-of-the-art LLM-based decoding, post-processing, and error-correcting methods across multiple datasets. Our approach proves particularly effective when processing challenging inputs such as complex speech sentences, acronyms, and domain-specific vocabulary.

eess.AS↗

arXiv2025

UmbraTTS: Adapting Text-to-Speech to Environmental Contexts with Flow Matching

Recent advances in Text-to-Speech (TTS) have enabled highly natural speech synthesis, yet integrating speech with complex background environments remains challenging. We introduce UmbraTTS, a flow-matching based TTS model that jointly generates both speech and environmental audio, conditioned on text and acoustic context. Our model allows fine-grained control over background volume and produces diverse, coherent, and context-aware audio scenes. A key challenge is the lack of data with speech and background audio aligned in natural context. To overcome the lack of paired training data, we propose a self-supervised framework that extracts speech, background audio, and transcripts from unannotated recordings. Extensive evaluations demonstrate that UmbraTTS significantly outperformed existing baselines, producing natural, high-quality, environmentally aware audios.

cs.SD↗

arXiv2025

FlowTSE: Target Speaker Extraction with Flow Matching

Target speaker extraction (TSE) aims to isolate a specific speaker's speech from a mixture using speaker enrollment as a reference. While most existing approaches are discriminative, recent generative methods for TSE achieve strong results. However, generative methods for TSE remain underexplored, with most existing approaches relying on complex pipelines and pretrained components, leading to computational overhead. In this work, we present FlowTSE, a simple yet effective TSE approach based on conditional flow matching. Our model receives an enrollment audio sample and a mixed speech signal, both represented as mel-spectrograms, with the objective of extracting the target speaker's clean speech. Furthermore, for tasks where phase reconstruction is crucial, we propose a novel vocoder conditioned on the complex STFT of the mixed signal, enabling improved phase estimation. Experimental results on standard TSE benchmarks show that FlowTSE matches or outperforms strong baselines.

eess.AS↗

arXiv2025

Spectral Analysis of Diffusion Models with Application to Schedule Design

Diffusion models (DMs) have emerged as powerful tools for modeling complex data distributions and generating realistic new samples. Over the years, advanced architectures and sampling methods have been developed to make these models practically usable. However, certain synthesis process decisions still rely on heuristics without a solid theoretical foundation. In our work, we offer a novel analysis of the DM's inference process, introducing a comprehensive frequency response perspective. Specifically, by relying on Gaussianity assumption, we present the inference process as a closed-form spectral transfer function, capturing how the generated signal evolves in response to the initial noise. We demonstrate how the proposed analysis can be leveraged to design a noise schedule that aligns effectively with the characteristics of the data. The spectral perspective also provides insights into the underlying dynamics and sheds light on the relationship between spectral properties and noise schedule structure. Our results lead to scheduling curves that are dependent on the spectral content of the data, offering a theoretical justification for some of the heuristics taken by practitioners.

cs.LG↗