mlsanityData to 14 Oct 2026

DSpAST: Disentangled Representations for Spatial Audio Reasoning with Large Language Models

arXiv 2509.13927 ↗First posted 17 Sep 2025eess.AS cs.AI cs.SDcited by 5

On the map: LLM reasoning → Audio-language model reasoning

Abstract (arXiv)

Reasoning about spatial audio with large language models requires a spatial audio encoder as an acoustic front-end to obtain audio embeddings for further processing. Such an encoder needs to capture all information required to detect the type of sound events, as well as the direction and distance of their corresponding sources. Accomplishing this with a single audio encoder is demanding as the information required for each of these tasks is mostly independent of each other. As a result, the performance obtained with a single encoder is often worse than when using task-specific audio encoders. In this work, we present DSpAST, a novel audio encoder based on SpatialAST that learns disentangled representations of spatial audio while having only 0.2% additional parameters. Experiments on SpatialSoundQA with the spatial audio reasoning system BAT demonstrate that DSpAST significantly outperforms SpatialAST.

Key facts. Cited by 5 papers in the map (data through 14 Oct 2026). Placed in Audio-language model reasoning, one of 17 directions in LLM reasoning. Builds on 30 of its 37 references that are in the map. First posted on arXiv 17 Sep 2025.

Builds on

30 of 37 references.

Closest ancestors · 2

Same direction, newest first

  1. 2025SAVVY: Spatial Awareness via Audio-Visual LLMs through Seeing and Hearing4 Jun
  2. 2024BAT: Learning to Reason about Spatial Sounds with Large Language Models2 Feb

Other references · 28

  1. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  2. 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,170
  3. 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 7,005
  4. 2018CBAM: Convolutional Block Attention Modulecited by 2,477
  5. 2020MiniLM: Deep Self-Attention Distillation for Task-Agnostic Compression of Pre-Trained Transformerscited by 1,463
  6. 2021AST: Audio Spectrogram Transformercited by 643
  7. 2022Masked Autoencoders that Listencited by 273
  8. 2023Listen, Think, and Understandcited by 209
  9. 2022SoundSpaces 2.0: A Simulation Platform for Visual-Acoustic Learningcited by 95
  10. 2019AdaCos: Adaptively Scaling Cosine Logits for Effectively Learning Deep Face Representationscited by 83
  11. 2023Whisper-AT: Noise-Robust Automatic Speech Recognizers are Also Strong General Audio Event Taggerscited by 72
  12. 2023Vision Transformer with Quadrangle Attentioncited by 46
16 more
  1. 2019Polyphonic Sound Event Detection and Localization using a Two-Stage Strategycited by 19
  2. 2024Can Large Language Models Understand Spatial Audio?cited by 15
  3. 2024Spatial Scaper: A Library to Simulate and Augment Soundscapes for Sound Event Localization and Detection in Realistic Roomscited by 12
  4. 2023Acoustic Prompt Tuning: Empowering Large Language Models with Audition Capabilitiescited by 10
  5. 2022Extending GCC-PHAT using Shift Equivariant Neural Networkscited by 7
  6. 2023Two vs. Four-Channel Sound Event Localization and Detectioncited by 6
  7. 2025Exploring Disentangled Neural Speech Codecs from Self-Supervised Representationscited by 3
  8. 2022Improved disentangled speech representations using contrastive learning in factorized hierarchical variational autoencodercited by 2
  9. 2024Speaker and Style Disentanglement of Speech Based on Contrastive Predictive Coding Supported Factorized Variational Autoencodercited by 2
  10. 2025Gen-A: Generalizing Ambisonics Neural Encoding to Unseen Microphone Arrayscited by 2
  11. 2025Baseline Systems and Evaluation Metrics for Spatial Semantic Segmentation of Sound Scenescited by 2
  12. 2025Description and Discussion on DCASE 2025 Challenge Task 4: Spatial Semantic Segmentation of Sound Scenescited by 2
  13. 2025Learning Robust Spatial Representations from Binaural Audio through Feature Distillationcited by 2
  14. 2022Binaural Signal Representations for Joint Sound Event Detection and Acoustic Scene Classificationcited by 1
  15. 2025Past, Present, and Future of Spatial Audio and Room Acousticscited by 1
  16. 2025Factorized RVQ-GAN For Disentangled Speech Tokenizationcited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 4.

5 papers.

  1. 2025SPUR: A Plug-and-Play Framework for Integrating Spatial Audio Understanding and Reasoning into Large Audio-Language Modelscited by 8
  2. 2026OmniEcho: Audio-Visual Spatial Understanding for Omni-Modal Embodied Agentscited by 2
  3. 2026Spectro-Temporal Interference Confounds Phase Encoding in Spatial Audio Foundation Modelscited by 0
  4. 2026SAIL: Spatial Audio Intelligence with Large Language Models via Disentangled Acoustic-Spatial Encoding and Dual-Stream Q-Formercited by 0
  5. 2026SEA-LM: Egocentric Spatial Audio Understanding for Wearable Microphone Arrayscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Spatial Audio Motion Understanding and Reasoning
  2. 2025OWL: Geometry-Aware Spatial Reasoning for Audio Large Language Models
  3. 2026The World is Not Mono: Enabling Spatial Understanding in Large Audio-Language Models
  4. 2026Spatial-Omni: Spatial Audio Understanding Integration in Multimodal LLMs via FOA Encoding
  5. 2026Spatial Audio Question Answering and Reasoning on Dynamic Source Movements
  6. 2025STAR-Bench: Probing Deep Spatio-Temporal Reasoning as Audio 4D Intelligence
  7. 2026SonicWorld: Source-Centric Latent Dynamics for Spatial Audio Reasoning
  8. 2026PhaseCoder: Microphone Geometry-Agnostic Spatial Audio Understanding for Multimodal LLMs