mlsanityData to 7 Oct 2026

NEUROTOKEN: Joint Source and Directional AAD with Envelope Decoding via Conditional Flow Matching

arXiv 2610.00397 ↗First posted 30 Sep 2026cs.LGcited by 0

On the map: Diffusion models → Audio, speech and music

Abstract (arXiv)

Identifying which speaker a listener is attending to in a noisy room -- the cocktail-party problem -- is the missing ingredient for next-generation hearing aids and brain-computer interfaces: it tells the device whose voice to amplify. Auditory attention decoding (AAD) reads this answer from EEG, but the literature splits into disconnected pieces: directional-AAD classifies side but does not map side to stream; regression-based source-AAD ranks candidate streams by a single Pearson correlation that is intrinsically noisy at the 1-5 s windows real devices need; and envelope reconstruction has no native AAD rule. We argue the right object is not any single statistic but the conditional likelihood of the attended envelope given EEG, and we make this practical with NEUROTOKEN: a single network whose three heads share one EEG front-end, with a conditional flow-matching head (ATTUNEFLOW) that scores candidates by an integrated velocity-residual likelihood ratio. Two inference-time ensembles -- QUADTRACK (four complementary statistics) and ENV-FLOW (z-normalised QUADTRACK+ATTUNEFLOW) -- absorb per-statistic failure modes for free. On KU Leuven, DTU, and NJU at 5 s, ATTUNEFLOW lifts per-segment source-AAD by 9%-16% over the strongest non-generative baseline and shrinks across-subject variance by ~3x; trial-level fusion exceeds 93% on two of three datasets. In parallel reproductions we show that canonical 95-97% direction-AAD numbers collapse by 17%-45% under a strict trial-disjoint protocol, clarifying both the true ceiling and why a likelihood-based formulation is needed.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Audio, speech and music, one of 14 directions in Diffusion models. Builds on 22 of its 41 references that are in the map. First posted on arXiv 30 Sep 2026.

Builds on

22 of 41 references.

References · 22

  1. 2018Representation Learning with Contrastive Predictive Codingcited by 7,662
  2. 2022Flow Matching for Generative Modelingcited by 5,426
  3. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flowcited by 3,550
  4. 2017FiLM: Visual Reasoning with a General Conditioning Layercited by 2,286
  5. 2023Voicebox: Text-Guided Multilingual Universal Speech Generation at Scalecited by 373
  6. 2024Large Brain Model for Learning Generic Representations with Tremendous EEG Data in BCIcited by 311
  7. 2022Decoding speech perception from non-invasive brain recordingscited by 129
  8. 2024WavTokenizer: an Efficient Acoustic Discrete Codec Tokenizer for Audio Language Modelingcited by 124
  9. 2024NeuroLM: A Universal Multi-task Foundation Model for Bridging the Gap between Language and EEG Signalscited by 83
  10. 2018Transfer Learning for Brain-Computer Interfaces: A Euclidean Space Data Alignment Approachcited by 76
  11. 2024SNAC: Multi-Scale Neural Audio Codeccited by 42
  12. 2024DARNet: Dual Attention Refinement Network with Spatiotemporal Construction for Auditory Attention Detectioncited by 10
10 more
  1. 2023A DenseNet-based method for decoding auditory spatial attention with EEGcited by 8
  2. 2025SSM2Mel: State Space Model to Reconstruct Mel Spectrogram from the EEGcited by 5
  3. 2023Auditory Attention Decoding with Task-Related Multi-View Contrastive Learningcited by 4
  4. 2024SWIM: Short-Window CNN Integrated with Mamba for EEG-Based Auditory Spatial Attention Decodingcited by 4
  5. 2024AADNet: An End-to-End Deep Learning Model for Auditory Attention Decodingcited by 4
  6. 2025DMF2Mel: A Dynamic Multiscale Fusion Network for EEG-Driven Mel Spectrogram Reconstructioncited by 3
  7. 2025ListenNet: A Lightweight Spatio-Temporal Enhancement Nested Network for Auditory Attention Detectioncited by 2
  8. 2025MHANet: Multi-scale Hybrid Attention Network for Auditory Attention Detectioncited by 2
  9. 2026DECAF: Dynamic Envelope Context-Aware Fusion for Speech-Envelope Reconstruction from EEGcited by 2
  10. 2024A contrastive-learning approach for auditory attention detectioncited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026RAMamba-Net: A Reliability-Aware and Mamba-Based Multimodal Fusion Network for Auditory Attention Detection
  2. 2026Addressing Limited Data in Auditory Attention Decoding with Diffusion Generative Models
  3. 2026NeuroSonic: Conditional Flow Matching for EEG-to-Speech Reconstruction
  4. 2025Improved Feature Extraction Network for Neuro-Oriented Target Speaker Extraction
  5. 2026Inside the Latent Flow: Causal Deciphering of Attention Dynamics in Audio Separation Foundation Models
  6. 2026Sensor Geometry as a Flow-Matching Prior for Multi-Channel Brain Signals
  7. 2025A Survey on Bridging EEG Signals and Generative AI: From Image and Text to Beyond
  8. 2026BrainStack: Neuro-MoE with Functionally Guided Expert Routing for EEG-Based Language Decoding