Pith. sign in

REVIEW 17 cited by

A-JEPA: Joint-Embedding Predictive Architecture Can Listen

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2311.15830 v3 pith:YADD6DH5 submitted 2023-11-27 cs.SD cs.CVeess.AS

A-JEPA: Joint-Embedding Predictive Architecture Can Listen

classification cs.SD cs.CVeess.AS
keywords audioa-jepaencodermaskingtargetarchitecturecontextcurriculum
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

This paper presents that the masked-modeling principle driving the success of large foundational vision models can be effectively applied to audio by making predictions in a latent space. We introduce Audio-based Joint-Embedding Predictive Architecture (A-JEPA), a simple extension method for self-supervised learning from the audio spectrum. Following the design of I-JEPA, our A-JEPA encodes visible audio spectrogram patches with a curriculum masking strategy via context encoder, and predicts the representations of regions sampled at well-designed locations. The target representations of those regions are extracted by the exponential moving average of context encoder, \emph{i.e.}, target encoder, on the whole spectrogram. We find it beneficial to transfer random block masking into time-frequency aware masking in a curriculum manner, considering the complexity of highly correlated in local time and frequency in audio spectrograms. To enhance contextual semantic understanding and robustness, we fine-tune the encoder with a regularized masking on target datasets, instead of input dropping or zero. Empirically, when built with Vision Transformers structure, we find A-JEPA to be highly scalable and sets new state-of-the-art performance on multiple audio and speech classification tasks, outperforming other recent models that use externally supervised pre-training.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 17 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. PairAlign: A Framework for Sequence Tokenization via Self-Alignment with Applications to Audio Tokenization

    cs.LG 2026-05 unverdicted novelty 7.0

    PairAlign learns compact audio token sequences via self-alignment of paired content views using an autoregressive decoder, achieving strong cross-view consistency and edit-distance preservation while reducing token co...

  2. PairAlign: A Framework for Sequence Tokenization via Self-Alignment with Applications to Audio Tokenization

    cs.LG 2026-05 unverdicted novelty 7.0

    PairAlign learns compact variable-length token sequences for audio via self-alignment on paired content-preserving views, achieving 55% fewer archive tokens than VQ while preserving edit-distance retrieval at 12.71 tokens/s.

  3. Recurrent Video Masked Autoencoders

    cs.CV 2025-12 unverdicted novelty 7.0

    RVM uses recurrent computation inside a masked autoencoder to learn video representations that match or exceed prior video and image models on classification, tracking, and dense spatial tasks with up to 30x better pa...

  4. The JEPA Paradox in Language: The Geometry of Linguistic Alternatives

    cs.CL 2026-07 conditional novelty 6.0

    Pure squared-error JEPA on masked text saturates mutual information and keeps high conditional variance, then collapses in rank and cosine similarity and transfers poorly, unlike matched I-JEPA on images.

  5. Music-JEPA: Learning a World Model of Sound from Action

    cs.SD 2026-07 conditional novelty 6.0

    An action-conditioned JEPA trained on paired piano audio and MIDI learns latent sound dynamics that support MIR tasks and transcription-style planning.

  6. Joint-Embedding Predictive Architecture for Sensor-based Activity Recognition

    eess.SP 2026-07 conditional novelty 6.0

    JEPA-based self-supervised pre-training on inertial sensor data improves recognition of rare transitional human activities compared to supervised learning, with gains mostly on transition classes.

  7. Frequency-Aware Self-Supervised Music Representation Learning

    cs.SD 2026-06 unverdicted novelty 6.0

    PupuJEPA applies a visual JEPA framework to 2D spectrograms with music-specific adaptations and outperforms 1D SSL models on the MARBLE benchmark for multiple MIR tasks.

  8. Frequency-Aware Self-Supervised Music Representation Learning

    cs.SD 2026-06 unverdicted novelty 6.0

    PupuJEPA applies a 2D joint-embedding predictive architecture to music spectrograms and outperforms 1D SSL models on the MARBLE benchmark.

  9. MJEPA: A Simple and Scalable Joint-Embedding Predictive Architecture for Audio-Visual Learning

    cs.CV 2026-06 unverdicted novelty 6.0

    MJ EPA applies a single shared ViT encoder and one predictive objective within and across audio-visual modalities, reporting >6.8 mAP gains on AudioSet-20K and competitive video results with 10x less data.

  10. AudioMosaic: Contrastive Masked Audio Representation Learning

    cs.LG 2026-05 unverdicted novelty 6.0

    AudioMosaic learns general-purpose audio representations through contrastive pre-training with structured spectrogram masking, reaching state-of-the-art results on standard benchmarks and improving audio-language tasks.

  11. AeroJEPA: Learning Semantic Latent Representations for Scalable 3D Aerodynamic Field Modeling

    cs.LG 2026-05 unverdicted novelty 6.0

    AeroJEPA applies joint-embedding predictive learning to produce scalable, semantically organized latent representations for 3D aerodynamic fields that support both field reconstruction and downstream design tasks.

  12. Self-Distillation of Hidden Layers for Self-Supervised Representation Learning

    cs.CV 2026-03 conditional novelty 6.0

    Predicting the outputs of several hidden layers of an EMA teacher, rather than only the final layer or pixels, substantially improves self-supervised ViT representations on ImageNet and downstream tasks.

  13. Joint-Embedding Predictive Architecture for Solar PV Panel Fault Classification

    eess.IV 2026-07 accept novelty 5.5

    JEFFNet fuses StoP-JEPA semantic embeddings with EfficientNetV2-S features for thermal IR PV fault classification, beating GEPFNet on F1 for multiclass and binary tasks with 47% fewer parameters.

  14. Scalable and Efficient Joint Spiking Embedding Predictive Architecture for Large-Scale Dynamic Graphs

    cs.LG 2026-07 conditional novelty 5.0

    SG-JEPA applies joint-embedding predictive learning to dynamic graphs, using spiking-neuron context encoders to predict future node embeddings without edge reconstruction or graph augmentation.

  15. AGE: Adaptive-masking for Graph Embedding in Graph Retrieval-Augmented Generation

    cs.IR 2026-06 unverdicted novelty 5.0

    AGE applies adaptive masking via a learnable sampler in Transformer-based SSL to align graph and text embeddings, yielding higher accuracy on four GraphQA benchmarks for non-parametric GraphRAG.

  16. Stylistic-STORM (ST-STORM) : Perceiving the Semantic Nature of Appearance

    cs.CV 2026-04 unverdicted novelty 5.0

    ST-STORM introduces a dual-branch SSL framework that disentangles semantic content from stylistic appearance using gated latent streams, JEPA for content invariance, and adversarial constraints for style capture.

  17. World Action Models: A Survey

    cs.RO 2026-06 unverdicted novelty 3.0

    A survey that clarifies boundaries and organizes World Action Models by generation requirements and predictive substrates, identifying a trend toward generating less of the future.