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

arXiv: https://arxiv.org/abs/2610.00397 · HTML: https://mlsanity.com/paper/2610.00397/

## Key facts

- Cited by 1 papers in the map (data through 14 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.

## Placed in

- Diffusion models / [Audio, speech and music](https://mlsanity.com/diffusion/audio-speech-and-music/)

## Builds on (most cited first, 22 of 22)

- [Representation Learning with Contrastive Predictive Coding](https://mlsanity.com/paper/1807.03748/) (2018)
- [Flow Matching for Generative Modeling](https://mlsanity.com/paper/2210.02747/) (2022)
- [Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow](https://mlsanity.com/paper/2209.03003/) (2022)
- [FiLM: Visual Reasoning with a General Conditioning Layer](https://mlsanity.com/paper/1709.07871/) (2017)
- [Voicebox: Text-Guided Multilingual Universal Speech Generation at Scale](https://mlsanity.com/paper/2306.15687/) (2023)
- [Large Brain Model for Learning Generic Representations with Tremendous EEG Data in BCI](https://mlsanity.com/paper/2405.18765/) (2024)
- [Decoding speech perception from non-invasive brain recordings](https://mlsanity.com/paper/2208.12266/) (2022)
- [WavTokenizer: an Efficient Acoustic Discrete Codec Tokenizer for Audio Language Modeling](https://mlsanity.com/paper/2408.16532/) (2024)
- [NeuroLM: A Universal Multi-task Foundation Model for Bridging the Gap between Language and EEG Signals](https://mlsanity.com/paper/2409.00101/) (2024)
- [Transfer Learning for Brain-Computer Interfaces: A Euclidean Space Data Alignment Approach](https://mlsanity.com/paper/1808.05464/) (2018)
- [SNAC: Multi-Scale Neural Audio Codec](https://mlsanity.com/paper/2410.14411/) (2024)
- [DARNet: Dual Attention Refinement Network with Spatiotemporal Construction for Auditory Attention Detection](https://mlsanity.com/paper/2410.11181/) (2024)
- [A DenseNet-based method for decoding auditory spatial attention with EEG](https://mlsanity.com/paper/2309.07690/) (2023)
- [SWIM: Short-Window CNN Integrated with Mamba for EEG-Based Auditory Spatial Attention Decoding](https://mlsanity.com/paper/2409.19884/) (2024)
- [SSM2Mel: State Space Model to Reconstruct Mel Spectrogram from the EEG](https://mlsanity.com/paper/2501.10402/) (2025)
- [Auditory Attention Decoding with Task-Related Multi-View Contrastive Learning](https://mlsanity.com/paper/2308.04244/) (2023)
- [AADNet: An End-to-End Deep Learning Model for Auditory Attention Decoding](https://mlsanity.com/paper/2410.13059/) (2024)
- [ListenNet: A Lightweight Spatio-Temporal Enhancement Nested Network for Auditory Attention Detection](https://mlsanity.com/paper/2505.10348/) (2025)
- [DMF2Mel: A Dynamic Multiscale Fusion Network for EEG-Driven Mel Spectrogram Reconstruction](https://mlsanity.com/paper/2507.07526/) (2025)
- [MHANet: Multi-scale Hybrid Attention Network for Auditory Attention Detection](https://mlsanity.com/paper/2505.15364/) (2025)
- [DECAF: Dynamic Envelope Context-Aware Fusion for Speech-Envelope Reconstruction from EEG](https://mlsanity.com/paper/2602.19395/) (2026)
- [A contrastive-learning approach for auditory attention detection](https://mlsanity.com/paper/2410.18395/) (2024)

## Built on by (most cited first, 1 of 1)

- [Merit: Making every modality earn its contribution in multimodal attention decoding](https://mlsanity.com/paper/2610.17507/) (2026)

## 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.