Audio Representation Learning

Follow this topic Ask about this topic

Audio representation learning studies how to encode waveforms or spectrograms into features that preserve information useful for tasks such as speech recognition, sound classification, music retrieval, and generation. Common approaches include self-supervised masked prediction in wav2vec 2.0, HuBERT, and BEST-RQ; contrastive audio-language learning such as CLAP; and distillation or autoencoder-based continuous and discrete audio tokenizers. Researchers assess transfer with linear probes and fine-tuning on benchmarks including SUPERB, ESC-50, MARBLE, and X-ARES, while also probing which acoustic, semantic, spatial, or speaker attributes the features retain. A central tension is how to make representations broadly transferable and compact while preserving fine acoustic detail, disentangling nuisance factors, and supporting both understanding and high-fidelity generation.

Latest papers

211 papers30 papers in the last four weeks, up 25% on the four weeks before. 0.2% of all new papers.

Mar 14, 2026

Evaluating Compositional Structure in Audio Representations

We propose a benchmark for evaluating compositionality in audio representations. Audio compositionality refers to representing sound scenes in terms of constituent sources and attributes, and combining them systematically. While central to auditory perception, this property is largely absent from current evaluation protocols. Our framework adapts ideas from vision and language to audio through two tasks: A-COAT, which tests consistency under additive transformations, and A-TRE, which probes reconstructibility from attribute-level primitives. Both tasks are supported by large synthetic datasets with controlled variation in acoustic attributes, providing the first benchmark of compositional structure in audio embeddings.

Chuyang Chen, Bea Steers, Brian McFee +1

Music and Audio Research Laboratory, New York University, USA

Compositional Representation Learning Audio Representation Learning

Mar 11, 2026

Fair-Gate: Fairness-Aware Interpretable Risk Gating for Sex-Fair Voice Biometrics

Voice biometric systems can exhibit sex-related performance gaps even when overall verification accuracy is strong. We attribute these gaps to two practical mechanisms: (i) demographic shortcut learning, where speaker classification training exploits spurious correlations between sex and speaker identity, and (ii) feature entanglement, where sex-linked acoustic variation overlaps with identity cues and cannot be removed without degrading speaker discrimination. We propose Fair-Gate, a fairness-aware and interpretable risk-gating framework that addresses both mechanisms in a single pipeline. Fair-Gate applies risk extrapolation to reduce variation in speaker-classification risk across proxy sex groups, and introduces a local complementary gate that routes intermediate features into an identity branch and a sex branch. The gate provides interpretability by producing an explicit routing mask that can be inspected to understand which features are allocated to identity versus sex-related pathways. Experiments on VoxCeleb1 show that Fair-Gate improves the utility--fairness trade-off, yielding more sex-fair ASV performance under challenging evaluation conditions.

Yangyang Qu, Massimiliano Todisco, Chiara Galdi +1

EURECOM Sophia Antipolis, France

Speaker Verification Algorithmic Fairness Audio Representation Learning Representation Disentanglement

Mar 11, 2026

Speaker Verification with Speech-Aware LLMs: Evaluation and Augmentation

Speech-aware large language models (LLMs) can accept speech inputs, yet their training objectives largely emphasize linguistic content or specific fields such as emotions or the speaker's gender, leaving it unclear whether they encode speaker identity. First, we propose a model-agnostic scoring protocol that produces continuous verification scores for both API-only and open-weight models, using confidence scores or log-likelihood ratios from the Yes/No token probabilities. Using this protocol, we benchmark recent speech-aware LLMs and observe weak speaker discrimination (EERs above 20% on VoxCeleb1). Second, we introduce a lightweight augmentation that equips an LLM with ASV capability by injecting frozen ECAPA-TDNN speaker embeddings through a learned projection and training only LoRA adapters. On TinyLLaMA-1.1B, the resulting ECAPA-LLM achieves 1.03% EER on VoxCeleb1-E, approaching a dedicated speaker verification system while preserving a natural-language interface.

Thomas Thebaud, Yuzhe Wang, Laureano Moro-Velazquez +2

Electrical and Computer Engineering Department, Johns Hopkins University, Baltimore, MD, USA · Human Language Technology Center of Excellence, Johns Hopkins University, Baltimore, MD, USA

Speaker Embedding Speaker Verification Speech Language Models Audio Representation Learning

Mar 11, 2026

Speech Codec Probing from Semantic and Phonetic Perspectives

Speech tokenizers are essential for connecting speech to large language models (LLMs) in multimodal systems. Speech tokenizers are expected to preserve both semantic and acoustic information for downstream understanding and generation tasks. However, emerging evidence suggests that the term "semantic" in speech processing does not align with linguistic lexical-semantic, leading to a mismatch between speech and text modality. In this paper, we systematically analyze the information encoded by several widely used speech tokenizers, evaluating their lexical-semantic and phonetic content through three tasks. Our results show that current tokenizers primarily capture phonetic rather than lexical-semantic structure, deriving practical implications for the design of next-generation speech tokenization methods. Code is released to public at https://github.com/Alexuan/codec_probing_release.

Xuan Shi, Chang Zeng, Tiantian Feng +3

University of Southern California, USA · Dolby Laboratories, USA

Audio Representation Learning Neural Audio Codecs Speech Processing Audio Tokenization

Mar 10, 2026

ALARM: Audio-Language Alignment for Reasoning Models

Large audio language models (ALMs) extend LLMs with auditory understanding. A common approach freezes the LLM and trains only an adapter on self-generated targets. However, this fails for reasoning LLMs (RLMs) whose built-in chain-of-thought traces expose the textual surrogate input, yielding unnatural responses. We propose self-rephrasing, converting self-generated responses into audio-understanding variants compatible with RLMs while preserving distributional alignment. We further fuse and compress multiple audio encoders for stronger representations. For training, we construct a 6M-instance multi-task corpus (2.5M unique prompts) spanning 19K hours of speech, music, and sound. Our 4B-parameter ALM outperforms similarly sized models and surpasses most larger ALMs on related audio-reasoning benchmarks, while preserving textual capabilities with a low training cost. Notably, we achieve the best open-source result on the MMAU-speech and MMSU benchmarks and rank third among all the models.

Petr Grinberg, Hassan Shahmohammadi

EPFL, Switzerland · Sony Europe Ltd., Germany

Audio-Language Models Audio Reasoning Audio Representation Learning Self-Training for Language Models

Mar 10, 2026

TimberAgent: Gram-Guided Retrieval for Executable Music Effect Control

Digital audio workstations expose rich effect chains, yet a semantic gap remains between perceptual user intent and low-level signal-processing parameters. We study retrieval-grounded audio effect control, where the output is an editable plugin configuration rather than a finalized waveform. Our focus is Texture Resonance Retrieval (TRR), an audio representation built from Gram matrices of projected mid-level Wav2Vec2 activations. This design preserves texture-relevant co-activation structure. We evaluate TRR on a guitar-effects benchmark with 1,063 candidate presets and 204 queries. The evaluation follows Protocol-A, a cross-validation scheme that prevents train-test leakage. We compare TRR against CLAP and internal retrieval baselines (Wav2Vec-RAG, Text-RAG, FeatureNN-RAG), using min-max normalized metrics grounded in physical DSP parameter ranges. Ablation studies validate TRR's core design choices: projection dimensionality, layer selection, and projection type. A near-duplicate sensitivity analysis confirms that results are robust to trivial knowledge-base matches. TRR achieves the lowest normalized parameter error among evaluated methods. A multiple-stimulus listening study with 26 participants provides complementary perceptual evidence. We interpret these results as benchmark evidence that texture-aware retrieval is useful for editable audio effect control, while broader personalization and real-audio robustness claims remain outside the verified evidence presented here.

Shihao He, Yihan Xia, Fang Liu +2

College of Electronics and Information Engineering, Shenzhen University, Shenzhen, China

Music Information Retrieval Audio Representation Learning

Mar 9, 2026

Universal Speech Content Factorization

We propose Universal Speech Content Factorization (USCF), a simple and invertible linear method for extracting a low-rank speech representation in which speaker timbre is suppressed while phonetic content is preserved. USCF extends Speech Content Factorization, a closed-set voice conversion (VC) method, to an open-set setting by learning a universal speech-to-content mapping via least-squares optimization and deriving speaker-specific transformations from only a few seconds of target speech. We show through embedding analysis that USCF effectively removes speaker-dependent variation. As a zero-shot VC system, USCF achieves competitive intelligibility, naturalness, and speaker similarity compared to methods that require substantially more target-speaker data or additional neural training. Finally, we demonstrate that as a training-efficient timbre-disentangled speech feature, USCF features can serve as the acoustic representation for training timbre-prompted text-to-speech models. Speech samples and code are publicly available.

Henry Li Xinyuan, Zexin Cai, Lin Zhang +5

Human Language Technology Center of Excellence (COE), Johns Hopkins University, USA · Center for Language and Speech Processing, Johns Hopkins University, USA

Voice Conversion Speaker Disentanglement Audio Representation Learning

Mar 9, 2026

Learning Multiple Utterance-Level Attribute Representations with a Unified Speech Encoder

Speech foundation models trained with self-supervised learning produce generic speech representations that support a wide range of speech processing tasks. When further adapted with supervised learning, these models can achieve strong performance on specific downstream tasks. Recent post-training approaches, such as SAMU-XSLR and SONAR, align speech representations with utterance-level semantic representations, enabling effective multimodal (speech-text) and multilingual applications. While speech foundation models typically learn contextual embeddings at the acoustic frame level, these methods learn representations at the utterance level. In this work, we extend this paradigm to arbitrary utterance-level attributes and propose a unified post-training framework that enables a single speech foundation model to generate multiple types of utterance-level representations. We demonstrate the effectiveness of this approach by jointly learning semantic and speaker representations and evaluating them on multilingual speech retrieval and speaker recognition tasks.

Maryem Bouziane, Salima Mdhaffar, Yannick Estève

LIA - Avignon Université, France

Speech Processing Audio Representation Learning Speech Foundation Models

Mar 9, 2026

SAVE: Speech-Aware Video Representation Learning for Video-Text Retrieval

For video-text retrieval, the use of CLIP has been a de facto choice. Since CLIP provides only image and text encoders, this consensus has led to a biased paradigm that entirely ignores the sound track of videos. While several attempts have been made to reintroduce audio -- typically by incorporating an audio encoder and fusing its output with visual features -- these methods face two challenges: ineffective representation of speech content and suboptimal vision-audio fusion. To address these issues jointly, we propose SAVE, a Speech Aware Video rEpresentation learning method. SAVE improves upon AVIGATE, a SOTA audiovisual method, with a dedicated speech branch for more effective speech embedding. Furthermore, we introduce soft-ALBEF for early vision-audio alignment that facilitates fusion. Extensive experiments on five benchmarks show that SAVE compares favorably against the SOTA, outperforming AVIGATE by +4.1% on MSRVTT-9k, +1.9% on MSRVTT-7k, +2.5% on VATEX, +9.8% on Charades, and +2.1% on LSMDC, in light of the SumR metric.

Ruixiang Zhao, Zhihao Xu, Bangxiang Lan +3

Renmin University of China

Audio-Visual Representation Learning Video Representation Learning Vision-Language Retrieval Audio-Visual Alignment Audio Representation Learning

Mar 5, 2026

Koopman Regularized Deep Speech Disentanglement for Speaker Verification

Human speech contains both linguistic content and speaker dependent characteristics making speaker verification a key technology in identity critical applications. Modern deep learning speaker verification systems aim to learn speaker representations that are invariant to semantic content and nuisance factors such as ambient noise. However, many existing approaches depend on labelled data, textual supervision or large pretrained models as feature extractors, limiting scalability and practical deployment, raising sustainability concerns. We propose Deep Koopman Speech Disentanglement Autoencoder (DKSD-AE), a structured autoencoder that combines a novel multi-step Koopman operator learning module with instance normalization to disentangle speaker and content dynamics. Quantitative experiments across multiple datasets demonstrate that DKSD-AE achieves improved or competitive speaker verification performance compared to state-of-the-art baselines while maintaining high content EER, confirming effective disentanglement. These results are obtained with substantially fewer parameters and without textual supervision. Moreover, performance remains stable under increased evaluation scale, highlighting representation robustness and generalization. Our findings suggest that Koopman-based temporal modelling, when combined with instance normalization, provides an efficient and principled solution for speaker-focused representation learning.

Nikos Chazaridis, Mohammad Belal, Rafael Mestre +2

School of Electronics and Computer Science, University of Southampton, SO17 1BJ Southampton, U.K. · Marine Geosciences, National Oceanography Centre, SO14 3ZH Southampton, U.K.

Speaker Verification Speaker Disentanglement Audio Representation Learning Disentangled Representation Learning Koopman Operator Learning

Mar 3, 2026

Interpreting Speaker Characteristics in the Dimensions of Self-Supervised Speech Features

How do speech models trained through self-supervised learning structure their representations? Previous studies have looked at how information is encoded in feature vectors across different layers. But few studies have considered whether speech characteristics are captured within individual dimensions of SSL features. In this paper, we specifically look at speaker information using PCA on utterance-averaged representations. For a range of SSL models, we find that the principal dimension that explains most variance encodes pitch and associated characteristics like gender. Other individual principal dimensions correlate with intensity, noise levels, the second formant, and higher frequency characteristics. We then use synthesis analyses to show that the dimensions for most characteristics are isolated from each other's influence. We further show that characteristics can be changed by manipulating the corresponding dimensions.

Kyle Janse van Rensburg, Benjamin van Niekerk, Herman Kamper

Department of Electrical and Electronic Engineering, Stellenbosch University, South Africa · Concordia University, Canada

Self-Supervised Speech Representation Learning Audio Representation Learning Disentangled Representation Learning

Mar 1, 2026

AG-REPA: Causal Layer Selection for Representation Alignment in Audio Flow Matching

REPresentation Alignment (REPA) improves the training of generative flow models by aligning intermediate hidden states with pretrained teacher features, but its effectiveness in token-conditioned audio Flow Matching critically depends on the choice of supervised layers, which is typically made heuristically based on the depth. In this work, we introduce Attribution-Guided REPresentation Alignment (AG-REPA), a novel causal layer selection strategy for representation alignment in audio Flow Matching. Firstly, we find that layers that best store semantic/acoustic information (high teacher-space similarity) are not necessarily the layers that contribute most to the velocity field that drives generation, and we call it Store-Contribute Dissociation (SCD). To turn this insight into an actionable training guidance, we propose a forward-only gate ablation (FoG-A) that quantifies each layer's causal contribution via the induced change in the predicted velocity field, enabling sparse layer selection and adaptive weighting for alignment. Across unified speech and general-audio training (LibriSpeech + AudioSet) under different token-conditioning topologies, AG-REPA consistently outperforms REPA baselines. Overall, our results show that alignment is most effective when applied to the causally dominant layers that drive the velocity field, rather than to layers that are representationally rich but functionally passive.

Pengfei Zhang, Tianxin Xie, Minghao Yang +1

AI Thrust, Information Hub, The Hong Kong University of Science and Technology (Guangzhou).

Flow-Based Generative Modeling Flow Matching Audio Representation Learning Feature Attribution

Feb 27, 2026

An Empirical Analysis of Task-Induced Encoder Bias in Fréchet Audio Distance

Fréchet Audio Distance (FAD) is the de facto standard for evaluating text-to-audio generation, yet its scores depend on the underlying encoder's embedding space. An encoder's training task dictates which acoustic features are preserved or discarded, causing FAD to inherit systematic task-induced biases. We decompose evaluation into Recall, Precision, and Alignment (split into semantic and structural dimensions), using log-scale normalization for fair cross-encoder comparison. Controlled experiments on six encoders across two datasets reveal a four-axis trade-off: reconstruction-based AudioMAE leads precision sensitivity; ASR-trained Whisper dominates structural detection but is blind to signal degradation; classification-trained VGGish maximizes semantic detection but penalizes legitimate intra-class variation. Since no single encoder is a universal evaluator, future metrics must shift toward evaluation-native encoders intrinsically aligned with human perception.

Wonwoo Jeong

Dept. of Computer Science and Engineering, Sogang University, South Korea

Text-to-Audio Generation Audio Representation Learning

Feb 27, 2026

DashengTokenizer: One layer is enough for unified audio understanding and generation

This paper introduces DashengTokenizer, a continuous audio tokenizer engineered for joint use in both understanding and generation tasks. Unlike conventional approaches, which train acoustic tokenizers and subsequently integrate frozen semantic knowledge, our method inverts this paradigm: we leverage frozen semantic features and inject acoustic information. In linear evaluation across 22 diverse tasks, our method outperforms previous audio codec and audio encoder baselines by a significant margin while maintaining competitive audio reconstruction quality. Notably, we demonstrate that this acoustic injection improves performance for tasks such as speech emotion recognition, music understanding, and acoustic scene classification. We further evaluate the tokenizer's generative performance on text-to-audio (TTA), text-to-music (TTM), and speech enhancement (SE). Our approach surpasses standard variational autoencoder (VAE)-based methods on TTA and TTM tasks, while its effectiveness on SE underscores its capabilities as a general-purpose audio encoder. Finally, our results challenge the prevailing assumption that VAE-based architectures are a prerequisite for audio synthesis. Checkpoints are available at https://huggingface.co/mispeech/dashengtokenizer.

Heinrich Dinkel, Xingwei Sun, Gang Li +8

MiLM Plus, Xiaomi Inc., Beijing, China

Audio Tokenization Audio Representation Learning Audio Understanding Text-to-Audio Generation Text-to-Music Generation

Feb 26, 2026

SemanticVocoder: Bridging Audio Generation and Audio Understanding via Semantic Latents

Recent audio generation models typically rely on Variational Autoencoders (VAEs) and perform generation within the VAE latent space. Although VAEs excel at compression and reconstruction, their latents inherently encode low-level acoustic details rather than semantically discriminative information, leading to entangled event semantics and complicating the training of generative models. To address these issues, we discard VAE acoustic latents and introduce semantic encoder latents, thereby proposing SemanticVocoder, a generative vocoder that directly synthesizes waveforms from semantic latents. Equipped with SemanticVocoder, our text-to-audio generation model achieves a Frechet Distance of 12.823 and a Frechet Audio Distance of 1.709 on the AudioCaps test set, as the introduced semantic latents exhibit superior discriminability compared to acoustic VAE latents. Beyond improved generation performance, it also serves as a promising attempt towards unifying audio understanding and generation within a shared semantic space. Generated samples are available at https://zeyuxie29.github.io/SemanticVocoder/.

Zeyu Xie, Chenxing Li, Qiao Jin +6

Peking University, Shenzhen · Tencent AI Lab, Shenzhen · Shanghai AI Lab, Shanghai +1

Neural Vocoders Text-to-Audio Generation Audio Representation Learning Audio Understanding

Feb 25, 2026

UniWhisper: Efficient Continual Multi-task Training for Robust Universal Audio Representation

A universal audio representation should capture fine-grained speech cues and high-level semantics for environmental sounds and music in a single encoder. Existing encoders often excel in one domain but degrade in others. We propose UniWhisper, an efficient continual multi-task training framework that casts heterogeneous audio tasks into a unified instruction and answer format. This enables standard next-token training without task-specific heads and losses. We train it on 38k hours of public audio and assess the encoder using shallow MLP probes and k-nearest neighbors (kNN) on 20 tasks spanning speech, environmental sound, and music. UniWhisper reaches normalized weighted averages of 0.81 with MLP probes and 0.61 with kNN, compared to 0.64 and 0.46 for Whisper, while retaining strong speech performance.

Yuxuan Chen, Peize He, Haoyuan Yu +1

Jilin University · University of Electronic Science and Technology of China · Hunan University +1

Audio Representation Learning Audio Understanding Multi-Task Learning Continual Learning

Feb 20, 2026

Musical Training, but not Mere Exposure to Music, Drives the Emergence of Chroma Equivalence in Artificial Neural Networks

Pitch is a fundamental aspect of auditory perception. Pitch perception is commonly described across two perceptual dimensions: pitch height is the sense that tones with varying frequencies seem to be higher or lower, and chroma equivalence is the cyclical similarity of notes octaves, corresponding to a doubling of fundamental frequency. Existing research is divided on whether chroma equivalence is a learned percept that varies according to musical experience and culture, or is an innate percept that develops automatically. Building on a recent framework that proposes to use ANNs to ask 'why' questions about the brain, we evaluated recent auditory ANNs using representational similarity analysis to test the emergence of pitch height and chroma equivalence in their learned representations. Additionally, we fine-tuned two models, Wav2Vec 2.0 and Data2Vec, on a self-supervised learning task using speech and music, and a supervised music transcription task. We found that all models exhibited varying degrees of pitch height representation, but that only models trained on the supervised music transcription task exhibited chroma equivalence. Mere exposure to music through self-supervised learning was not sufficient for chroma equivalence to emerge. This supports the view that chroma equivalence is a higher-order cognitive computation that emerges to support the specific task of music perception, distinct from other auditory perception such as speech listening. This work also highlights the usefulness of ANNs for probing the developmental conditions that give rise to perceptual representations in humans.

Lukas Grasse, Matthew S. Tata

Canadian Centre for Behavioural Neuroscience, Department of Neuroscience, University of Lethbridge, AB, Canada, T1K 3Y4

Audio Representation Learning

Feb 18, 2026

BAT: Better Audio Transformer Guided by Convex Gated Probing

Probing is widely adopted in computer vision to faithfully evaluate self-supervised learning (SSL) embeddings, as finetuning may misrepresent their inherent quality. In contrast, audio SSL models still rely on finetuning because simple probing fails to unlock their full potential and alters their rankings when competing on AudioSet. Hence, a robust and efficient probing mechanism is required to guide the trajectory of audio SSL towards reliable and reproducible methods. We introduce Convex Gated Probing (CGP), a prototype-based method that significantly closes the gap between finetuning and probing in audio. CGP efficiently utilizes all frozen layers via a gating mechanism and exposes the location of latent task-relevant information. Guided by CGP as a reliable post-hoc evaluation probe, we rework the entire SSL pipeline of current best performing audio models that use legacy implementations of prior SSL methods. By refining data preprocessing, model architecture, and pretraining recipe, we introduce Better Audio Transformer (BAT), and establish new SOTA on audio benchmarks.

Houtan Ghaffari, Lukas Rauch, Christoph Scholz +1

Ghent University · University of Kassel · Now at Earth Species Project +1

Audio Representation Learning Self-Supervised Pre-Training

Feb 17, 2026

A Generative-First Neural Audio Autoencoder

Neural autoencoders underpin generative models. Practical, large-scale use of neural autoencoders for generative modeling necessitates fast encoding, low latent rates, and a single model across representations. Existing approaches are reconstruction-first: they incur high latent rates, slow encoding, and separate architectures for discrete vs. continuous latents and for different audio channel formats, hindering workflows from preprocessing to inference conditioning. We introduce a generative-first architecture for audio autoencoding that increases temporal downsampling from 2048x to 3360x and supports continuous and discrete representations and common audio channel formats in one model. By balancing compression, quality, and speed, it delivers 10x faster encoding, 1.6x lower rates, and eliminates channel-format-specific variants while maintaining competitive reconstruction quality. This enables applications previously constrained by processing costs: a 60-second mono signal compresses to 788 tokens, making generative modeling more tractable.

Jonah Casebeer, Ge Zhu, Zhepei Wang +1

Adobe Research

Autoencoders Audio Representation Learning Neural Audio Codecs Audio Token Compression

Feb 17, 2026

What Do Neurons Listen To? A Neuron-level Dissection of a General-purpose Audio Model

In this paper, we analyze the internal representations of a general-purpose audio self-supervised learning (SSL) model from a neuron-level perspective. Despite their strong empirical performance as feature extractors, the internal mechanisms underlying the robust generalization of SSL audio models remain unclear. Drawing on the framework of mechanistic interpretability, we identify and examine class-specific neurons by analyzing conditional activation patterns across diverse tasks. Our analysis reveals that SSL models foster the emergence of class-specific neurons that provide extensive coverage across novel task classes. These neurons exhibit shared responses across different semantic categories and acoustic similarities, such as speech attributes and musical pitch. We also confirm that these neurons have a functional impact on classification performance. To our knowledge, this is the first systematic neuron-level analysis of a general-purpose audio SSL model, providing new insights into its internal representation.

Takao Kawamura, Daisuke Niizumi, Nobutaka Ono

Graduate School of Systems Design, Tokyo Metropolitan University, Tokyo, Japan

Audio Representation Learning Neural Network Interpretability Self-Supervised Learning

Jan 16, 2026

Scalable Music Cover Retrieval Using Lyrics-Aligned Audio Embeddings

Music Cover Retrieval, also known as Version Identification, aims to recognize distinct renditions of the same underlying musical work, a task central to catalog management, copyright enforcement, and music retrieval. State-of-the-art approaches have largely focused on harmonic and melodic features, employing increasingly complex audio pipelines designed to be invariant to musical attributes that often vary widely across covers. While effective, these methods demand substantial training time and computational resources. By contrast, lyrics constitute a strong invariant across covers, though their use has been limited by the difficulty of extracting them accurately and efficiently from polyphonic audio. Early methods relied on simple frameworks that limited downstream performance, while more recent systems deliver stronger results but require large models integrated within complex multimodal architectures. We introduce LIVI (Lyrics-Informed Version Identification), an approach that seeks to balance retrieval accuracy with computational efficiency. First, LIVI leverages supervision from state-of-the-art transcription and text embedding models during training to achieve retrieval accuracy on par with--or superior to--harmonic-based systems. Second, LIVI remains lightweight and efficient by removing the transcription step at inference, challenging the dominance of complexity-heavy pipelines.

Joanne Affolter, Benjamin Martin, Elena V. Epure +2

Deezer Research - 24 Rue de Calais, 75009 Paris · EPFL DHLAB - Rte Cantonale, 1015 Lausanne

Audio Representation Learning Music Information Retrieval

Jan 14, 2026

SITA: Learning Speaker-Invariant and Tone-Aware Speech Representations for Low-Resource Tonal Languages

Tonal low-resource languages are widely spoken but remain underserved by modern speech technologies. A central challenge is learning speech representations that are robust to nuisance variation, such as speaker gender, while preserving lexical tone, which carries word meaning. We propose SITA, a lightweight adaptation recipe for pretrained wav2vec-style self-supervised speech encoders. Rather than designing a new backbone or objective, SITA combines existing objectives in a staged optimization framework to reduce tone collapse while preserving ASR capability. Stage 1 improves speaker invariance without erasing tonal contrasts by combining a cross-gender contrastive loss with a tone-repulsive loss that separates same-word, different-tone realizations. Stage 2 restores recognition-oriented linguistic information through CTC fine-tuning and knowledge distillation on upper encoder layers. We evaluate SITA primarily on Hmong, a tonal language with limited digital resources and a small speaker pool. Against multilingual, speaker-adversarial, label-aware, and semi-supervised baselines, SITA achieves the best trade-off between cross-gender lexical retrieval and tone separation, while maintaining ASR accuracy close to an ASR-adapted XLS-R teacher. Results on Mandarin show consistent gains, suggesting that SITA is a general plug-in recipe for tonal speech representation learning.

Tianyi Xu, Xuan Ouyang, Binwei Yao +4

University of Wisconsin–Madison, Madison, WI, USA

Audio Representation Learning Self-Supervised Speech Representation Learning Low-Resource Speech Recognition

Jan 13, 2026

Linear probing enables Ship-Radiated Noise recognition with pretrained audio embeddings

Even though the ocean covers the majority of the planet's surface, it remains the least explored ecosystem. As light and radio waves do not propagate through water, underwater acoustics is the main choice for various ocean applications ranging from marine biology to pollution monitoring. Increasing levels of anthropogenic noise from ships contribute significantly to underwater sound pollution, posing risks to marine ecosystems. This makes monitoring crucial to understand and quantify the impact of the ship radiated noise. Passive Acoustic Monitoring (PAM) systems are widely deployed for this purpose, generating years of underwater recordings across diverse soundscapes. Manual analysis of such large-scale data is impractical, motivating the need for automated approaches based on machine learning. Recent advances in automatic Underwater Acoustic Target Recognition (UATR) have largely relied on supervised learning, which is constrained by the scarcity of labeled data. Transfer Learning (TL) offers a promising alternative to mitigate this limitation. In this work, we conduct the first empirical comparative study of transfer learning for UATR, evaluating multiple pretrained audio models originating from diverse audio domains. The pretrained model weights are frozen, and the resulting embeddings are analyzed through classification, clustering, and similarity-based evaluations. The analysis shows that the geometrical structure of the embedding space is largely dominated by recording-specific characteristics. However, a simple linear probe can effectively suppress this recording-specific information and isolate ship-type features from these embeddings. As a result, linear probing enables effective automatic UATR using pretrained audio models at low computational cost, significantly reducing the need for a large amounts of high-quality labeled ship recordings.

Hilde I. Hummel, Sandjai Bhulai, Rob D. van der Mei +1

Centre of Mathematics and Computer Science, Stochastics group, Science Park 123, Amsterdam, 1098 XG, Netherlands · Vrije Universiteit, Department of Mathematics, De Boelelaan 1111, Amsterdam, 1081 HV, Netherlands · Naturalis Biodiversity Center, Darwinweg 2, Leiden, 2333 CR, Netherlands

Audio Representation Learning Linear Probing Underwater Acoustics Transfer Learning Passive Acoustic Monitoring

Jan 11, 2026

Categorize Early, Integrate Late: Divergent Processing Strategies in Automatic Speech Recognition

In speech language modeling, two architectures dominate the frontier: the Transformer and the Conformer. However, it remains unknown whether their comparable performance stems from convergent processing strategies or distinct architectural inductive biases. We introduce Architectural Fingerprinting, a probing framework that isolates the effect of architecture on representation, and apply it to a controlled suite of 24 pre-trained encoders (39M-3.3B parameters). Our analysis reveals divergent hierarchies: Conformers implement a "Categorize Early" strategy, resolving phoneme categories 29% earlier in depth and speaker gender by 16% depth. In contrast, Transformers "Integrate Late," deferring phoneme, accent, and duration encoding to deep layers (49-57%). These fingerprints suggest design heuristics: Conformers' front-loaded categorization may benefit low-latency streaming, while Transformers' deep integration may favor tasks requiring rich context and cross-utterance normalization.

Nathan Roll, Pranav Bhalerao, Martijn Bartelds +7

Stanford University

Audio Representation Learning Hybrid CNN-Transformer Architectures Transformer Automatic Speech Recognition Representation Probing

Dec 12, 2025

The Affective Bridge: Preserving Speech Representations while Enhancing Deepfake Detection via emotional Constraints

Speech deepfake detection (DFD) has benefited from diverse acoustic and semantic speech representations, many of which encode valuable speech information and are costly to train. Prior work has shown that affective cues improve DFD, yet existing approaches either fuse emotion with other task-specific features in complex pipelines or directly fine-tune representations toward DFD objectives, risking distortion of the original speech representations that support downstream tasks such as speaker verification (SV) or automatic speech recognition (ASR). We propose a simpler approach: fine-tuning speech encoders on emotion recognition alone-without any DFD supervision, and training a lightweight support vector machine (SVM) on the frozen emotion-tuned representations for DFD. This preserves the original representation capacity for downstream tasks such as SV and ASR, while emergently improving DFD performance. Crucially, we find that emotion is uniquely effective as this bridging task: replacing it with speaker identity even degrades DFD performance, demonstrating that the benefit stems from emotion's role as a natural bridge between speech representation and DFD. Experiments on FakeOrReal and In-the-Wild show accuracy improvements of up to 6% and 2% with corresponding EER reductions, while analysis on ASVspoof 2019 LA reveals dataset-specific speaker bias in the real-speech subset. Code is available at GitHub repository.

Yupei Li, Chenyang Lyu, Longyue Wang +3

Audio Deepfake Detection Audio Representation Learning

Nov 26, 2025

Harmonic-Percussive Disentangled Neural Audio Codec for Bandwidth Extension

Bandwidth extension, the task of reconstructing the high-frequency components of an audio signal from its low-passed counterpart, is a long-standing problem in audio processing. In this work, we extend recent advances in neural architectures by framing bandwidth extension as an audio token prediction problem. Specifically, we train a transformer-based language model on the discrete representations produced by a disentangled neural audio codec, where the disentanglement is guided by a Harmonic-Percussive decomposition of the input signals, highlighting spectral structures particularly relevant for bandwidth extension. Our approach introduces a novel codec design that explicitly accounts for the downstream token prediction task, enabling a more effective coupling between codec structure and transformer modeling. This joint design yields high-quality reconstructions of the original signal, as measured by both objective metrics and subjective evaluations. These results highlight the importance of aligning codec disentanglement and representation learning with the generative modeling stage, and demonstrate the potential of global, representation-aware design for advancing bandwidth extension.

Benoît Giniès, Xiaoyu Bie, Olivier Fercoq +1

LTCI, Télécom Paris, Institut Polytechnique de Paris Palaiseau, France

Audio Representation Learning Neural Audio Codecs

Nov 20, 2025

Revisiting Audio-language Pretraining for Learning General-purpose Audio Representation

Audio-language pretraining (ALP) holds promise for learning general-purpose audio representation, yet remains underexplored. Crucially, there is no consensus on whether audio-language models can build effective general-purpose audio encoders, nor a systematic understanding of how pretraining objectives behave across diverse tasks and scales. We identify three key barriers: limited scale of audio-text corpora, limited coverage of audio attributes in existing caption corpora, and lack of systematic exploration and evaluation. To fill this gap, we present the first principled empirical study of ALP. We first introduce CaptionStew, a 10.7M caption dataset aggregating open-source audio-text corpora across multiple domains and captioning focuses. We then conduct the first comprehensive evaluation comparing contrastive and captioning objectives for learning audio representation across speech, music, and environmental sound tasks. Our results not only demonstrate that ALP yields competitive, transferable representations, but reveal critical trade-offs: contrastive learning offers superior data efficiency, while captioning exhibits better scalability. Furthermore, we find that the benefits of supervised initialization often diminish at larger scales, challenging common practices. By grounding these claims in empirical evidence, we establish a viable pathway toward general-purpose audio representation learning, guiding future research.

Wei-Cheng Tseng, Xuanru Zhou, Mingyue Huo +3

University of Texas at Austin · Tencent AI Lab Seattle · Zhejiang University +1

Audio-Language Model Evaluation Audio Representation Learning Audio Understanding Audio-Language Models Audio Captioning Multimodal Contrastive Learning

Nov 7, 2025

Perceptually Aligning Representations of Music via Noise-Augmented Autoencoders

We argue that training autoencoders to reconstruct inputs from noised versions of their encodings, when combined with perceptually motivated losses, yields encodings that are structured according to a perceptual hierarchy. We demonstrate the emergence of this hierarchy by showing that, after training an audio autoencoder in this manner, perceptually salient information is captured in coarser representation structures than with conventional training. Furthermore, we show that such perceptual hierarchies improve latent diffusion decoding in the context of estimating pitch surprisal in music and predicting EEG-brain responses to music listening. In both cases, our results surpass those of previous methods. Pretrained weights are available on github.com/CPJKU/pa-audioic.

Mathias Rose Bjare, Giorgia Cantisani, Marco Pasini +2

Johannes Kepler University, Linz, AT · STMS, CNRS, IRCAM, Sorbonne Universit´e, Paris, FR · Queen Mary University, London, UK +1

Audio Representation Learning Denoising Autoencoders Autoencoders Hierarchical Representation Learning

Jul 18, 2025

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

Shikhar Bharadwaj, Samuele Cornell, Kwanghee Choi +4

Carnegie Mellon University, USA · National Institute of Advanced Industrial Science and Technology (AIST), Japan

Audio Representation Learning Audio QA Audio Understanding Bioacoustics

Mar 20, 2025

Structured-Noise Masked Modeling for Video, Audio and Beyond

Masked modeling has emerged as a robust self-supervised learning framework. However, most methods rely on random masking, which disregards the structural properties of different data modalities. To align with the spatiotemporal and spectral characteristics of video and audio data, we introduce a structured noise-based masking approach. By filtering white noise into different color noise distributions, we generate structured masks that capture modality-specific patterns without requiring handcrafted heuristics or access to the data. Our approach enhances masked video and audio modeling frameworks without any additional computational cost. Experiments show that structured noise masking consistently outperforms random masking, underscoring the value of modality-aware masking strategies for representation learning.

Aritra Bhowmik, Carlos Hinojosa, Fida Mohammad Thoker +2

University of Amsterdam · King Abdullah University of Science and Technology

Audio Representation Learning Video Representation Learning