arXiv subjects

Paris Smaragdis

Publications and source records attributed to Paris Smaragdis.

At least 19 recordsLinked to original sources

arXiv2026

Elastic Time: Dynamic Frame Rate Bottlenecks for Neural Audio Coding

Neural audio autoencoders have become a core component of compression, feature extraction, and generation. However, while existing systems support variable bitrate, the vast majority of models still operate at a fixed latent frame-rate, allocating equal temporal budget to regions with very different information density, which can result in unnecessarily long sequences. We introduce Elastic Time, a dynamic frame-rate bottleneck that converts fixed-frame-rate autoencoders to dynamic ones. Our method learns a lightweight latent predictor used to decide which frames can be skipped and later reconstructed, enabling efficient greedy boundary selection at inference. Experiments show our method enables deployment-time rate control while improving efficiency-quality tradeoffs relative to baselines. Overall, we provide a flexible mechanism for adjusting temporal resolution in audio autoencoders, potentially facilitating more efficient downstream modeling for generation and long-context tasks.

cs.SD↗

arXiv2026

Diagnosing Neural Convergence with Topological Alignment Spectra

Representational similarity in neural networks is inherently scale-dependent, yet widely used metrics such as Centered Kernel Alignment (CKA) and Procrustes analysis provide only global scalar estimates. These scalars often fail to distinguish micro-scale geometric jitter (local noise) from macro-scale semantic reorganization, compressing multi-scale structural relationships into a single uninformative value. We introduce the Topological Alignment Spectrum (TAS), a multi-scale diagnostic tool that sweeps normalized mean Jaccard similarity over varying neighborhood sizes. By normalizing the metric over an analytically-derived expected range (from expected overlap under randomness to perfect alignment), TAS yields a dimension-invariant metric over a spectrum of scales, where one indicates perfect structural alignment, zero reflects chance-level agreement, and negative values signal active anti-alignment at specific scales. Experiments on synthetic point clouds demonstrate that TAS allows the recognition of distinct types of alignment perturbation: local jitter harms fine-grained neighborhoods but preserves cluster-level structure, while cluster-center shuffling preserves local similarity but disrupts global alignment -- phenomena that remain invisible or conflated under global, single-scalar metrics. Applying TAS to the MultiBERTs collection reveals that fine-tuning induces comprehensive topological reorganization across scales, challenging the view of task adaptation as merely conservative or localized. While models from different random seeds remain locally divergent, semantic clusters emerge as the dominant scale of alignment. TAS thus offers a granular, topology-aware alternative for diagnosing convergence and representational stability in deep networks.

cs.LG↗

arXiv2026

Gencho: Room Impulse Response Generation from Reverberant Speech and Text via Diffusion Transformers

Blind room impulse response (RIR) estimation is a core task for capturing and transferring acoustic properties; yet existing methods often suffer from limited modeling capability and degraded performance under unseen conditions. Moreover, emerging generative audio applications call for more flexible impulse response generation methods. We propose Gencho, a diffusion-transformer-based model that predicts complex spectrogram RIRs from reverberant speech. A structure-aware encoder leverages isolation between early and late reflections to encode the input audio into a robust representation for conditioning, while the diffusion decoder generates diverse and perceptually realistic impulse responses from it. Gencho integrates modularly with standard speech processing pipelines for acoustic matching. Results show richer generated RIRs than non-generative baselines while maintaining strong performance in standard RIR metrics. We further demonstrate its application to text-conditioned RIR generation, highlighting Gencho's versatility for controllable acoustic simulation and generative audio tasks.

cs.SD↗

arXiv2025

PromptSep: Generative Audio Separation via Multimodal Prompting

Recent breakthroughs in language-queried audio source separation (LASS) have shown that generative models can achieve higher separation audio quality than traditional masking-based approaches. However, two key limitations restrict their practical use: (1) users often require operations beyond separation, such as sound removal; and (2) relying solely on text prompts can be unintuitive for specifying sound sources. In this paper, we propose PromptSep to extend LASS into a broader framework for general-purpose sound separation. PromptSep leverages a conditional diffusion model enhanced with elaborated data simulation to enable both audio extraction and sound removal. To move beyond text-only queries, we incorporate vocal imitation as an additional and more intuitive conditioning modality for our model, by incorporating Sketch2Sound as a data augmentation strategy. Both objective and subjective evaluations on multiple benchmarks demonstrate that PromptSep achieves state-of-the-art performance in sound removal and vocal-imitation-guided source separation, while maintaining competitive results on language-queried source separation.

cs.SD↗

arXiv2025

Re-Bottleneck: Latent Re-Structuring for Neural Audio Autoencoders

Neural audio codecs and autoencoders have emerged as versatile models for audio compression, transmission, feature-extraction, and latent-space generation. However, a key limitation is that most are trained to maximize reconstruction fidelity, often neglecting the specific latent structure necessary for optimal performance in diverse downstream applications. We propose a simple, post-hoc framework to address this by modifying the bottleneck of a pre-trained autoencoder. Our method introduces a "Re-Bottleneck", an inner bottleneck trained exclusively through latent space losses to instill user-defined structure. We demonstrate the framework's effectiveness in three experiments. First, we enforce an ordering on latent channels without sacrificing reconstruction quality. Second, we align latents with semantic embeddings, analyzing the impact on downstream diffusion modeling. Third, we introduce equivariance, ensuring that a filtering operation on the input waveform directly corresponds to a specific transformation in the latent space. Ultimately, our Re-Bottleneck framework offers a flexible and efficient way to tailor representations of neural audio models, enabling them to seamlessly meet the varied demands of different applications with minimal additional training.

cs.SD↗

arXiv2025

Learning to Upsample and Upmix Audio in the Latent Domain

Neural audio autoencoders create compact latent representations that preserve perceptually important information, serving as the foundation for both modern audio compression systems and generation approaches like next-token prediction and latent diffusion. Despite their prevalence, most audio processing operations, such as spatial and spectral up-sampling, still inefficiently operate on raw waveforms or spectral representations rather than directly on these compressed representations. We propose a framework that performs audio processing operations entirely within an autoencoder's latent space, eliminating the need to decode to raw audio formats. Our approach dramatically simplifies training by operating solely in the latent domain, with a latent L1 reconstruction term, augmented by a single latent adversarial discriminator. This contrasts sharply with raw-audio methods that typically require complex combinations of multi-scale losses and discriminators. Through experiments in bandwidth extension and mono-to-stereo up-mixing, we demonstrate computational efficiency gains of up to 100x while maintaining quality comparable to post-processing on raw audio. This work establishes a more efficient paradigm for audio processing pipelines that already incorporate autoencoders, enabling significantly faster and more resource-efficient workflows across various audio tasks.

cs.SD↗

arXiv2025

Combolutional Neural Networks

Selecting appropriate inductive biases is an essential step in the design of machine learning models, especially when working with audio, where even short clips may contain millions of samples. To this end, we propose the combolutional layer: a learned-delay IIR comb filter and fused envelope detector, which extracts harmonic features in the time domain. We demonstrate the efficacy of the combolutional layer on three information retrieval tasks, evaluate its computational cost relative to other audio frontends, and provide efficient implementations for training. We find that the combolutional layer is an effective replacement for convolutional layers in audio tasks where precise harmonic analysis is important, e.g., piano transcription, speaker classification, and key detection. Additionally, the combolutional layer has several other key benefits over existing frontends, namely: low parameter count, efficient CPU inference, strictly real-valued computations, and improved interpretability.

cs.SD↗

arXiv2025

Continuous-Time Signal Decomposition: An Implicit Neural Generalization of PCA and ICA

We generalize the low-rank decomposition problem, such as principal and independent component analysis (PCA, ICA) for continuous-time vector-valued signals and provide a model-agnostic implicit neural signal representation framework to learn numerical approximations to solve the problem. Modeling signals as continuous-time stochastic processes, we unify the approaches to both the PCA and ICA problems in the continuous setting through a contrast function term in the network loss, enforcing the desired statistical properties of the source signals (decorrelation, independence) learned in the decomposition. This extension to a continuous domain allows the application of such decompositions to point clouds and irregularly sampled signals where standard techniques are not applicable.

cs.LG↗

arXiv2025

Rethinking Non-Negative Matrix Factorization with Implicit Neural Representations

Non-negative Matrix Factorization (NMF) is a powerful technique for analyzing regularly-sampled data, i.e., data that can be stored in a matrix. For audio, this has led to numerous applications using time-frequency (TF) representations like the Short-Time Fourier Transform. However extending these applications to irregularly-spaced TF representations, like the Constant-Q transform, wavelets, or sinusoidal analysis models, has not been possible since these representations cannot be directly stored in matrix form. In this paper, we formulate NMF in terms of learnable functions (instead of vectors) and show that NMF can be extended to a wider variety of signal classes that need not be regularly sampled.

eess.AS↗

arXiv2025

Adaptive Slimming for Scalable and Efficient Speech Enhancement

Speech enhancement (SE) enables robust speech recognition, real-time communication, hearing aids, and other applications where speech quality is crucial. However, deploying such systems on resource-constrained devices involves choosing a static trade-off between performance and computational efficiency. In this paper, we introduce dynamic slimming to DEMUCS, a popular SE architecture, making it scalable and input-adaptive. Slimming lets the model operate at different utilization factors (UF), each corresponding to a different performance/efficiency trade-off, effectively mimicking multiple model sizes without the extra storage costs. In addition, a router subnet, trained end-to-end with the backbone, determines the optimal UF for the current input. Thus, the system saves resources by adaptively selecting smaller UFs when additional complexity is unnecessary. We show that our solution is Pareto-optimal against individual UFs, confirming the benefits of dynamic routing. When training the proposed dynamically-slimmable model to use 10% of its capacity on average, we obtain the same or better speech quality as the equivalent static 25% utilization while reducing MACs by 29%.

eess.AS↗

arXiv2025

User-guided Generative Source Separation

Music source separation (MSS) aims to extract individual instrument sources from their mixture. While most existing methods focus on the widely adopted four-stem separation setup (vocals, bass, drums, and other instruments), this approach lacks the flexibility needed for real-world applications. To address this, we propose GuideSep, a diffusion-based MSS model capable of instrument-agnostic separation beyond the four-stem setup. GuideSep is conditioned on multiple inputs: a waveform mimicry condition, which can be easily provided by humming or playing the target melody, and mel-spectrogram domain masks, which offer additional guidance for separation. Unlike prior approaches that relied on fixed class labels or sound queries, our conditioning scheme, coupled with the generative approach, provides greater flexibility and applicability. Additionally, we design a mask-prediction baseline using the same model architecture to systematically compare predictive and generative approaches. Our objective and subjective evaluations demonstrate that GuideSep achieves high-quality separation while enabling more versatile instrument extraction, highlighting the potential of user participation in the diffusion-based generative process for MSS. Our code and demo page are available at https://yutongwen.github.io/GuideSep/

cs.SD↗

arXiv2025

Generative Data Augmentation Challenge: Synthesis of Room Acoustics for Speaker Distance Estimation

This paper describes the synthesis of the room acoustics challenge as a part of the generative data augmentation workshop at ICASSP 2025. The challenge defines a unique generative task that is designed to improve the quantity and diversity of the room impulse responses dataset so that it can be used for spatially sensitive downstream tasks: speaker distance estimation. The challenge identifies the technical difficulty in measuring or simulating many rooms' acoustic characteristics precisely. As a solution, it proposes generative data augmentation as an alternative that can potentially be used to improve various downstream tasks. The challenge website, dataset, and evaluation code are available at https://sites.google.com/view/genda2025.

eess.AS↗

arXiv2024

Audio Editing with Non-Rigid Text Prompts

In this paper, we explore audio-editing with non-rigid text edits. We show that the proposed editing pipeline is able to create audio edits that remain faithful to the input audio. We explore text prompts that perform addition, style transfer, and in-painting. We quantitatively and qualitatively show that the edits are able to obtain results which outperform Audio-LDM, a recently released text-prompted audio generation model. Qualitative inspection of the results points out that the edits given by our approach remain more faithful to the input audio in terms of keeping the original onsets and offsets of the audio events.

cs.SD↗

arXiv2024

On Class Separability Pitfalls In Audio-Text Contrastive Zero-Shot Learning

Recent advances in audio-text cross-modal contrastive learning have shown its potential towards zero-shot learning. One possibility for this is by projecting item embeddings from pre-trained backbone neural networks into a cross-modal space in which item similarity can be calculated in either domain. This process relies on a strong unimodal pre-training of the backbone networks, and on a data-intensive training task for the projectors. These two processes can be biased by unintentional data leakage, which can arise from using supervised learning in pre-training or from inadvertently training the cross-modal projection using labels from the zero-shot learning evaluation. In this study, we show that a significant part of the measured zero-shot learning accuracy is due to strengths inherited from the audio and text backbones, that is, they are not learned in the cross-modal domain and are not transferred from one modality to another.

cs.SD↗

arXiv2024

Scaling Up Adaptive Filter Optimizers

We introduce a new online adaptive filtering method called supervised multi-step adaptive filters (SMS-AF). Our method uses neural networks to control or optimize linear multi-delay or multi-channel frequency-domain filters and can flexibly scale-up performance at the cost of increased compute -- a property rarely addressed in the AF literature, but critical for many applications. To do so, we extend recent work with a set of improvements including feature pruning, a supervised loss, and multiple optimization steps per time-frame. These improvements work in a cohesive manner to unlock scaling. Furthermore, we show how our method relates to Kalman filtering and meta-adaptive filtering, making it seamlessly applicable to a diverse set of AF tasks. We evaluate our method on acoustic echo cancellation (AEC) and multi-channel speech enhancement tasks and compare against several baselines on standard synthetic and real-world datasets. Results show our method performance scales with inference cost and model capacity, yields multi-dB performance gains for both tasks, and is real-time capable on a single CPU core.

cs.SD↗

arXiv2024

Sound Source Separation Using Latent Variational Block-Wise Disentanglement

While neural network approaches have made significant strides in resolving classical signal processing problems, it is often the case that hybrid approaches that draw insight from both signal processing and neural networks produce more complete solutions. In this paper, we present a hybrid classical digital signal processing/deep neural network (DSP/DNN) approach to source separation (SS) highlighting the theoretical link between variational autoencoder and classical approaches to SS. We propose a system that transforms the single channel under-determined SS task to an equivalent multichannel over-determined SS problem in a properly designed latent space. The separation task in the latent space is treated as finding a variational block-wise disentangled representation of the mixture. We show empirically, that the design choices and the variational formulation of the task at hand motivated by the classical signal processing theoretical results lead to robustness to unseen out-of-distribution data and reduction of the overfitting risk. To address the resulting permutation issue we explicitly incorporate a novel differentiable permutation loss function and augment the model with a memory mechanism to keep track of the statistics of the individual sources.

eess.AS↗

arXiv2024

Noise-Robust DSP-Assisted Neural Pitch Estimation with Very Low Complexity

Pitch estimation is an essential step of many speech processing algorithms, including speech coding, synthesis, and enhancement. Recently, pitch estimators based on deep neural networks (DNNs) have have been outperforming well-established DSP-based techniques. Unfortunately, these new estimators can be impractical to deploy in real-time systems, both because of their relatively high complexity, and the fact that some require significant lookahead. We show that a hybrid estimator using a small deep neural network (DNN) with traditional DSP-based features can match or exceed the performance of pure DNN-based models, with a complexity and algorithmic delay comparable to traditional DSP-based algorithms. We further demonstrate that this hybrid approach can provide benefits for a neural vocoding task.

eess.AS↗

arXiv2023

Meta-AF Echo Cancellation for Improved Keyword Spotting

Adaptive filters (AFs) are vital for enhancing the performance of downstream tasks, such as speech recognition, sound event detection, and keyword spotting. However, traditional AF design prioritizes isolated signal-level objectives, often overlooking downstream task performance. This can lead to suboptimal performance. Recent research has leveraged meta-learning to automatically learn AF update rules from data, alleviating the need for manual tuning when using simple signal-level objectives. This paper improves the Meta-AF framework by expanding it to support end-to-end training for arbitrary downstream tasks. We focus on classification tasks, where we introduce a novel training methodology that harnesses self-supervision and classifier feedback. We evaluate our approach on the combined task of acoustic echo cancellation and keyword spotting. Our findings demonstrate consistent performance improvements with both pre-trained and joint-trained keyword spotting models across synthetic and real playback. Notably, these improvements come without requiring additional tuning, increased inference-time complexity, or reliance on oracle signal-level training data.

cs.SD↗