Finite Scalar Quantization Enables Redundant and Transmission-Robust Neural Audio Compression at Low Bit-rates

Neural Audio Codecs (NACs) have become increasingly adopted in speech processing tasks due to their excellent rate-distortion performance and compatibility with Large Language Models (LLMs) as discrete feature representations for audio generation. While most existing codecs rely on Residual Vector Quantization (RVQ), Finite Scalar Quantization (FSQ) has recently emerged as a compelling alternative that simplifies training and natively supports single codebooks. We introduce NeuCodec, an FSQ-based NAC, and show that FSQ encodes baked-in redundancy which produces an encoding which is robust when transmitted through noisy channels. First, through an encoder distillation experiment, we show that two different encoders can learn to encode identical audio into vastly different code sequences whilst maintaining comparable reconstruction quality with the same quantizer and decoder. Second, we demonstrate that FSQ has vastly superior bit-level perturbation robustness by comparing the performance of RVQ and FSQ codecs when simulating the transmission of code sequences through a noisy channel.

Acoustic to Articulatory Speech Inversion for Children with Velopharyngeal Insufficiency

Traditional clinical approaches for assessing nasality, such as nasopharyngoscopy and nasometry, involve unpleasant experiences and are problematic for children. Speech Inversion (SI), a noninvasive technique, offers a promising alternative for estimating articulatory movement without the need for physical instrumentation. In this study, an SI system trained on nasalance data from healthy adults is augmented with source information from electroglottography and acoustically derived F0, periodic and aperiodic energy estimates as proxies for glottal control. This model achieves 16.92% relative improvement in Pearson Product-Moment Correlation (PPMC) compared to a previous SI system for nasalance estimation. To adapt the SI system for nasalance estimation in children with Velopharyngeal Insufficiency (VPI), the model initially trained on adult speech was fine-tuned using children with VPI data, yielding an 7.90% relative improvement in PPMC compared to its performance before fine-tuning.

Listening for "You": Enhancing Speech Image Retrieval via Target Speaker Extraction

Image retrieval using spoken language cues has emerged as a promising direction in multimodal perception, yet leveraging speech in multi-speaker scenarios remains challenging. We propose a novel Target Speaker Speech-Image Retrieval task and a framework that learns the relationship between images and multi-speaker speech signals in the presence of a target speaker. Our method integrates pre-trained self-supervised audio encoders with vision models via target speaker-aware contrastive learning, conditioned on a Target Speaker Extraction and Retrieval module. This enables the system to extract spoken commands from the target speaker and align them with corresponding images. Experiments on SpokenCOCO2Mix and SpokenCOCO3Mix show that TSRE significantly outperforms existing methods, achieving 36.3% and 29.9% Recall@1 in 2 and 3 speaker scenarios, respectively - substantial improvements over single speaker baselines and state-of-the-art models. Our approach demonstrates potential for real-world deployment in assistive robotics and multimodal interaction systems.

MAPSS: Manifold-based Assessment of Perceptual Source Separation

Objective assessment of source-separation systems still mismatches subjective human perception, especially when leakage and self-distortion interact. We introduce the Perceptual Separation (PS) and Perceptual Match (PM), the first pair of measures that functionally isolate these two factors. Our intrusive method begins with generating a bank of fundamental distortions for each reference waveform signal in the mixture. Distortions, references, and their respective system outputs from all sources are then independently encoded by a pre-trained self-supervised learning model. These representations are aggregated and projected onto a manifold via diffusion maps, which aligns Euclidean distances on the manifold with dissimilarities of the encoded waveforms. On this manifold, the PM measures the Mahalanobis distance from each output to its attributed cluster that consists of its reference and distortions embeddings, capturing self-distortion. The PS accounts for the Mahalanobis distance of the output to the attributed and to the closest non-attributed clusters, quantifying leakage. Both measures are differentiable and granular, operating at a resolution as low as 50 frames per second. We further derive, for both measures, deterministic error radius and non-asymptotic, high-probability confidence intervals (CIs). Experiments on English, Spanish, and music mixtures show that the PS and PM nearly always achieve the highest linear correlation coefficients with human mean-opinion scores than 14 competitors, reaching as high as 86.36% for speech and 87.21% for music. We observe, at worst, an error radius of 1.39% and a probabilistic 95% CI of 12.21% for these coefficients, which improves reliable and informed evaluation. Using mutual information, the measures complement each other most as their values decrease, suggesting they are jointly more informative as system performance degrades.

GmSLM : Generative Marmoset Spoken Language Modeling

Marmoset monkeys exhibit complex vocal communication, challenging the view that nonhuman primates vocal communication is entirely innate, and show similar features of human speech, such as vocal labeling of others and turn-taking. Studying their vocal communication offers a unique opportunity to link it with brain activity-especially given the difficulty of accessing the human brain in speech and language research. Since Marmosets communicate primarily through vocalizations, applying standard LLM approaches is not straightforward. We introduce Generative Marmoset Spoken Language Modeling (GmSLM), an optimized spoken language model pipeline for Marmoset vocal communication. We designed a novel zero-shot evaluation metrics using unsupervised in-the-wild data, alongside weakly labeled conversational data, to assess GmSLM and demonstrate its advantage over a basic human-speech-based baseline. GmSLM generated vocalizations closely matched real resynthesized samples acoustically and performed well on downstream tasks. Despite being fully unsupervised, GmSLM effectively distinguish real from artificial conversations and may support further investigations of the neural basis of vocal communication and provides a practical framework linking vocalization and brain activity. We believe GmSLM stands to benefit future work in neuroscience, bioacoustics, and evolutionary biology. Samples are provided under: pages.cs.huji.ac.il/adiyoss-lab/GmSLM.

EchoX: Towards Mitigating Acoustic-Semantic Gap via Echo Training for Speech-to-Speech LLMs

Speech-to-speech large language models (SLLMs) are attracting increasing attention. Derived from text-based large language models (LLMs), SLLMs often exhibit degradation in knowledge and reasoning capabilities. We hypothesize that this limitation arises because current training paradigms for SLLMs fail to bridge the acoustic-semantic gap in the feature representation space. To address this issue, we propose EchoX, which leverages semantic representations and dynamically generates speech training targets. This approach integrates both acoustic and semantic learning, enabling EchoX to preserve strong reasoning abilities as a speech LLM. Experimental results demonstrate that EchoX, with about six thousand hours of training data, achieves advanced performance on multiple knowledge-based question-answering benchmarks. The project is available at https://github.com/FreedomIntelligence/EchoX.

LITcoder: A General-Purpose Library for Building and Comparing Encoding Models

We introduce LITcoder, an open-source library for building and benchmarking neural encoding models. Designed as a flexible backend, LITcoder provides standardized tools for aligning continuous stimuli (e.g., text and speech) with brain data, transforming stimuli into representational features, mapping those features onto brain data, and evaluating the predictive performance of the resulting model on held-out data. The library implements a modular pipeline covering a wide array of methodological design choices, so researchers can easily compose, compare, and extend encoding models without reinventing core infrastructure. Such choices include brain datasets, brain regions, stimulus feature (both neural-net-based and control, such as word rate), downsampling approaches, and many others. In addition, the library provides built-in logging, plotting, and seamless integration with experiment tracking platforms such as Weights & Biases (W&B). We demonstrate the scalability and versatility of our framework by fitting a range of encoding models to three story listening datasets: LeBel et al. (2023), Narratives, and Little Prince. We also explore the methodological choices critical for building encoding models for continuous fMRI data, illustrating the importance of accounting for all tokens in a TR scan (as opposed to just taking the last one, even when contextualized), incorporating hemodynamic lag effects, using train-test splits that minimize information leakage, and accounting for head motion effects on encoding model predictivity. Overall, LITcoder lowers technical barriers to encoding model implementation, facilitates systematic comparisons across models and datasets, fosters methodological rigor, and accelerates the development of high-quality high-performance predictive models of brain activity. Project page: https://litcoder-brain.github.io

DiFlow-TTS: Discrete Flow Matching with Factorized Speech Tokens for Low-Latency Zero-Shot Text-To-Speech

Zero-shot Text-to-Speech (TTS) aims to synthesize high-quality speech that mimics the voice of an unseen speaker using only a short reference sample, requiring not only speaker adaptation but also accurate modeling of prosodic attributes. Recent approaches based on language models, diffusion, and flow matching have shown promising results in zero-shot TTS, but still suffer from slow inference and repetition artifacts. Discrete codec representations have been widely adopted for speech synthesis, and recent works have begun to explore diffusion models in purely discrete settings, suggesting the potential of discrete generative modeling for speech synthesis. However, existing flow-matching methods typically embed these discrete tokens into a continuous space and apply continuous flow matching, which may not fully leverage the advantages of discrete representations. To address these challenges, we introduce DiFlow-TTS, which, to the best of our knowledge, is the first model to explore purely Discrete Flow Matching for speech synthesis. DiFlow-TTS explicitly models factorized speech attributes within a compact and unified architecture. It leverages in-context learning by conditioning on textual content, along with prosodic and acoustic attributes extracted from a reference speech, enabling effective attribute cloning in a zero-shot setting. In addition, the model employs a factorized flow prediction mechanism with distinct heads for prosody and acoustic details, allowing it to learn aspect-specific distributions. Experimental results demonstrate that DiFlow-TTS achieves promising performance in several key metrics, including naturalness, prosody, preservation of speaker style, and energy control. It also maintains a compact model size and achieves low-latency inference, generating speech up to 25.8 times faster than the latest existing baselines.

Behind the Scenes: Mechanistic Interpretability of LoRA-adapted Whisper for Speech Emotion Recognition

Large pre-trained speech models such as Whisper offer strong generalization but pose significant challenges for resource-efficient adaptation. Low-Rank Adaptation (LoRA) has become a popular parameter-efficient fine-tuning method, yet its underlying mechanisms in speech tasks remain poorly understood. In this work, we conduct the first systematic mechanistic interpretability study of LoRA within the Whisper encoder for speech emotion recognition (SER). Using a suite of analytical tools, including layer contribution probing, logit-lens inspection, and representational similarity via singular value decomposition (SVD) and centered kernel alignment (CKA), we reveal two key mechanisms: a delayed specialization process that preserves general features in early layers before consolidating task-specific information, and a forward alignment, backward differentiation dynamic between LoRA's matrices. Our findings clarify how LoRA reshapes encoder hierarchies, providing both empirical insights and a deeper mechanistic understanding for designing efficient and interpretable adaptation strategies in large speech models. Our code is available at https://github.com/harryporry77/Behind-the-Scenes.

HISPASpoof: A New Dataset For Spanish Speech Forensics

Zero-shot Voice Cloning (VC) and Text-to-Speech (TTS) methods have advanced rapidly, enabling the generation of highly realistic synthetic speech and raising serious concerns about their misuse. While numerous detectors have been developed for English and Chinese, Spanish-spoken by over 600 million people worldwide-remains underrepresented in speech forensics. To address this gap, we introduce HISPASpoof, the first large-scale Spanish dataset designed for synthetic speech detection and attribution. It includes real speech from public corpora across six accents and synthetic speech generated with six zero-shot TTS systems. We evaluate five representative methods, showing that detectors trained on English fail to generalize to Spanish, while training on HISPASpoof substantially improves detection. We also evaluate synthetic speech attribution performance on HISPASpoof, i.e., identifying the generation method of synthetic speech. HISPASpoof thus provides a critical benchmark for advancing reliable and inclusive speech forensics in Spanish.