Pith. sign in

REVIEW 3 major objections 3 minor 58 references

How Would It Sound? Material-Controlled Multimodal Acoustic Profile Generation for Indoor Scenes

T0 review · 3 major / 3 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read Given a single audio-visual view of an indoor scene and a user-defined material mask, the model generates the Room Impulse Response those materials would produce, without a 3D mesh or measured room acoustics.

desk verdict The task framing is genuinely new, but the provided text stops before the method or any quantitative results, so the performance claims are unverifiable from what we have; the full paper deserves peer review. read the letter →

arxiv 2508.02905 v1 pith:XUREV62V submitted 2025-08-04 cs.CV cs.SDeess.AS

classification cs.CVcs.SDeess.AS
keywords roomimpulseresponsematerial-awareacousticsaudio-visuallearningindoorsceneunderstandingacousticprofilegenerationcounterfactualsimulatorbenchmarkinteractivematerialcontrol
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper's proposal is a new task: take one audio-visual observation of an indoor scene and a user-supplied material configuration, and generate the Room Impulse Response the scene would have if its surfaces and objects were actually made of those materials. This matters because sound in a room depends heavily on materials, yet current RIR prediction either requires expensive 3D meshes and measurements or infers materials implicitly from images, so nobody can ask 'what if I put acoustic tiles here?' and get an answer. To make the task concrete, the paper builds a new simulator-based benchmark (Acoustic Wonderland Dataset) spanning room geometries, object layouts, and material sets, and trains an encoder-decoder that conditions on the material mask. The paper's reported results say the model encodes material information and produces higher-fidelity RIRs than existing baselines, with a user study suggesting it transfers toward real rooms.

What carries the argument

The load-bearing pieces are a material mask and an encoder-decoder that treats it as a conditioning signal. A material mask is a user-defined assignment of materials to the surfaces and objects of the scene; the model encodes the scene's visual and acoustic state from a single location, fuses it with the mask, and decodes a Room Impulse Response, the function that describes how sound from a source reaches a receiver after reflections, absorption, and reverberation. The benchmark is the other load-bearing component: simulated scenes provide paired observations, material masks, and ground-truth RIRs, including held-out configurations that test whether the model generalizes to materials and room shapes it has not seen.

What would settle it

Find or design a room where the dominant acoustic feature (an open corridor, a large absorber, a glass facade) is not visible in the provided view but is covered by the material mask; simulate or measure its true RIR under two material configurations, and compare the model's predicted RIRs to the measured difference. If the model's predicted difference is negligible or spectrally wrong while measured reverberation changes substantially, the single-view sufficiency assumption is false.

Watch

Extended reading notes

Core claim

The paper's central claim is that material information can be encoded explicitly and used at inference time to synthesize a plausible Room Impulse Response for a scene the user has not actually modified. The authors argue that existing RIR prediction methods either require a full 3D mesh or treat materials implicitly through RGB images, and that none lets the user assign new materials on the fly. Their proposed encoder-decoder accepts an audio-visual observation of a scene plus a user-defined material mask, and produces the RIR that would result from those materials. To evaluate this, the paper introduces the Acoustic Wonderland Dataset, a simulator-based benchmark with varied room geometries, object layouts, and material configurations, and reports that their model outperforms several baselines and existing methods, with a user study suggesting reasonable transfer to real-world conditions.

Load-bearing premise

The load-bearing premise is that a single audio-visual observation plus the user's material mask captures all the surface and geometry information needed to predict the room impulse response, including surfaces that are occluded or outside the field of view.

Editorial extensions

If this is right

  • A designer, architect, or VR/AR developer can audition material changes such as carpet instead of wood or acoustic tiles instead of concrete without altering the physical space or building a 3D mesh.
  • The Acoustic Wonderland Dataset provides a reproducible benchmark for material-aware RIR prediction, with splits for seen and unseen materials and geometries.
  • Because the conditioning is a mask, the model can generate multiple RIRs from the same scene by changing only the mask, enabling rapid comparisons of material configurations.
  • The reported generalization to unseen materials and geometries suggests the learned mapping is not merely memorizing training scenes.
  • The user study indicates the gap between simulated training and real-world listening may be small enough for practical use.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same conditioning pattern could be applied to other acoustic or environmental parameters, such as source and receiver positions, object density, or air absorption, since the model already separates scene encoding from user-controllable inputs.
  • The single-view assumption is likely to be most strained when the acoustically dominant surface is occluded or outside the frame; a useful stress test would compare predictions against a full-geometry simulator on scenes with deliberately hidden absorbers or openings.
  • A natural next benchmark would measure whether material awareness also improves downstream tasks such as speech dereverberation or audio-visual navigation, rather than only RIR fidelity.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 3 minor

Summary. The paper introduces a new task, material-controlled acoustic profile generation, in which an indoor scene is observed through audio-visual data and a user-specified material mask is used to generate a target Room Impulse Response (RIR) at inference time. The authors propose an encoder-decoder architecture that fuses the scene observation with the material configuration, and they describe a new benchmark, the Acoustic Wonderland Dataset, built with SoundSpaces 2.0 and Habitat 2.0. The abstract claims that the model outperforms several baselines and state-of-the-art methods and that a user study demonstrates real-world generalization.

Significance. If fully substantiated, the task is novel and practically valuable for AR/VR, acoustic design, and architectural evaluation, since it promises controllable material editing without requiring 3D meshes or measured acoustics. The proposed benchmark could also be a useful resource for the community. However, as submitted, the manuscript contains no method section, no dataset statistics, no training details, no evaluation tables, and no user-study results, so the central claims are entirely unsupported. The paper's contribution is therefore significant in intention but not verifiable in its present form.

major comments (3)
  1. [Entire manuscript] The manuscript is incomplete: after Section 2 (Related Works) the text jumps directly to the references, with no Section 3 describing the proposed method, no dataset section, no experimental setup, no results tables, and no user study. The abstract claims that the model 'outperforms several baselines and state-of-the-art methods' and that a user study 'demonstrates the model's ability to generalize well to real-world scenarios,' but none of these claims can be checked because the supporting content is absent. This is load-bearing: without the method and experiments, the paper's central contribution cannot be evaluated. The authors must provide the complete methodological and experimental content before the manuscript can be considered further.
  2. [Section 2 (Related Works)] The paper states that the method relies on 'multimodal observations from a single location, alleviating the need for scene meshes or explicit geometric properties.' This raises an identifiability concern: a single-view observation plus an image-space material mask does not, in general, determine the full set of room surfaces, occluded geometry, or off-view materials, yet the RIR depends on all such surfaces through early reflections and the late reverberant tail. The manuscript provides no arguments or experiments showing that the model can recover the necessary hidden information on novel scenes, e.g., when the wall behind the camera has acoustic treatment or when an open doorway leads to a different room. This is a correctness-risk concern that needs to be addressed, either by narrowing the task formulation to settings where single-view observations are sufficient or by adding ablations that explicitly vary occluded and off-view surface materials.
  3. [Abstract / benchmark description] The benchmark is described as using audio-visual simulators, with SoundSpaces 2.0 and Habitat 2.0 cited, but the paper does not specify whether the training and evaluation RIRs are all generated by the same simulator. If both are simulation-generated, the evaluation only measures how well the model imitates the simulator's material model, not real-world acoustics. The claimed user study is the only evidence for real-world generalization, and it is entirely absent from the submitted text. The authors should report quantitative cross-domain evaluation on measured RIRs (e.g., from existing measured datasets) or, at minimum, provide full details of the user study and any sim-to-real comparison.
minor comments (3)
  1. [Title page / author block] The author names are presented as a single run-on line ('Mahnoor Fatima Saad Ziad Al-Halah') with a single email address for two distinct identifiers. The authors should clarify the actual author names and affiliations.
  2. [Figure 1] Figure 1 contains garbled, unrelated text and graphics, including phrases such as '1st Flyby', 'State Variable: Last Epoch', 'Optimal Control Problem Multi-Stage Decision Problem', and corrupted strings like 'DOO *ODVV7LOH'. These appear to be remnants from another figure and should be replaced with a clean version of the intended illustration.
  3. [References] References [42] and [43] list the same paper title ('Acoustic classification and optimization for multi-modal rendering of real-world scenes') with different years and page ranges; this appears to be a duplicate entry and should be consolidated or corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No equation-level circularity: the material mask is an input, the target RIR comes from an independent simulator, and no visible equation or fitted parameter reduces the prediction to its inputs.

full rationale

The claimed derivation chain is: (audio-visual observation, user-defined material mask) -> learned encoder-decoder -> target RIR, with ground-truth RIRs rendered by SoundSpaces 2.0. I find no circular step in the available text. The material configuration is a conditioning input, not a fitted parameter, and the target RIR is produced by an independent acoustic simulator rather than by the model, so no prediction is equivalent to its input by construction. Self-citations to prior work by co-author Ziad Al-Halah (SoundSpaces 2.0 [8], Few-Shot Audio-Visual Learning [27], etc.) serve as simulator infrastructure, related work, and benchmark construction, not as an unverified uniqueness theorem that forces the model choice; they are independent, code-reproduced tools. The lack of an external measured-RIR benchmark is a genuine validation/generalizability limitation, but it is not circularity: it concerns whether the simulator's material model is a trustworthy ground truth, not whether the paper's output reduces to its input. Because the provided manuscript omits the method and experiment sections, no further equation-level audit is possible; based on the text available, the derivation is self-contained and no circular step can be substantiated.

Assumptions & free parameters 1 free parameters · 2 assumptions · 0 invented entities

The central claim rests on the fidelity of the simulator and on the sufficiency of single-view observations. No explicit free parameters or invented entities are named in the available text; all learned weights are standard fitted model parameters.

free parameters (1)
  • Neural network weights (encoder and decoder)
    Learned end-to-end on the simulated dataset; specific values are not reported in the available text. These are standard model parameters, not hand-chosen constants.
assumptions (2)
  • domain assumption Synthetic RIRs from SoundSpaces 2.0 are sufficiently accurate for learning material-dependent room acoustics that transfer to real scenes.
    The dataset and ground truth are generated with a simulator; the model's ability to generalize to real-world settings depends on this fidelity.
  • domain assumption A single-view audio-visual observation plus a material mask is sufficient to determine the room's acoustic response.
    The method avoids 3D geometry, so occluded surfaces and off-view geometry must be inferred from one observation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of How Would It Sound? Material-Controlled Multimodal Acoustic Profile Generation for Indoor Scenes." pith.science (2026). https://pith.science/paper/XUREV62V

@misc{pith2026250802905,
  author       = {Pith},
  title        = {Pith review of: How Would It Sound? Material-Controlled Multimodal Acoustic Profile Generation for Indoor Scenes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XUREV62V}},
  note         = {Machine review of arXiv:2508.02905}
}
read the original abstract

How would the sound in a studio change with a carpeted floor and acoustic tiles on the walls? We introduce the task of material-controlled acoustic profile generation, where, given an indoor scene with specific audio-visual characteristics, the goal is to generate a target acoustic profile based on a user-defined material configuration at inference time. We address this task with a novel encoder-decoder approach that encodes the scene's key properties from an audio-visual observation and generates the target Room Impulse Response (RIR) conditioned on the material specifications provided by the user. Our model enables the generation of diverse RIRs based on various material configurations defined dynamically at inference time. To support this task, we create a new benchmark, the Acoustic Wonderland Dataset, designed for developing and evaluating material-aware RIR prediction methods under diverse and challenging settings. Our results demonstrate that the proposed model effectively encodes material information and generates high-fidelity RIRs, outperforming several baselines and state-of-the-art methods.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

58 extracted references · 55 canonical work pages

  1. [1]

    Allen and David A

    Jont B. Allen and David A. Berkley. Image method for effi- ciently simulating small-room acoustics. The Journal of the Acoustical Society of America, 65(4):943–950, 1979. 2

  2. [2]

    Direct-to-Indirect Acoustic Radiance Transfer

    Lakulish Antani, Anish Chandak, Micah Taylor, and Dinesh Manocha. Direct-to-Indirect Acoustic Radiance Transfer. IEEE Transactions on Visualization and Computer Graphics (TVCG), 18(2):261–269, 2012. 2

  3. [3]

    Interactive sound propagation with bidirec- tional path tracing

    Chunxiao Cao, Zhong Ren, Carl Schissler, Dinesh Manocha, and Kun Zhou. Interactive sound propagation with bidirec- tional path tracing. ACM Transactions on Graphics (TOG), 35(6), 2016. 2

  4. [4]

    Matterport3D: Learning from RGB- D Data in Indoor Environments

    Angel Chang, Angela Dai, Thomas Funkhouser, Maciej Hal- ber, Matthias Niessner, Manolis Savva, Shuran Song, Andy Zeng, and Yinda Zhang. Matterport3D: Learning from RGB- D Data in Indoor Environments. International Conference on 3D Vision (3DV), 2017. 3

  5. [5]

    SoundSpaces: Audio- Visual Navigaton in 3D Environments

    Changan Chen, Unnat Jain, Carl Schissler, Sebastia Vi- cenc Amengual Gari, Ziad Al-Halah, Vamsi Krishna Ithapu, Philip Robinson, and Kristen Grauman. SoundSpaces: Audio- Visual Navigaton in 3D Environments. In Proceedings of the European Conference on Computer Vision (ECCV), 2020. 2

  6. [6]

    Seman- tic audio-visual navigation

    Changan Chen, Ziad Al-Halah, and Kristen Grauman. Seman- tic audio-visual navigation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 15516–15525, 2021. 2

  7. [7]

    Visual Acoustic Matching

    Changan Chen, Ruohan Gao, Paul Calamia, and Kristen Grauman. Visual Acoustic Matching. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 18858–18868, 2022. 3

  8. [8]

    SoundSpaces 2.0: A Simulation Platform for Visual-Acoustic Learning

    Changan Chen, Carl Schissler, Sanchit Garg, Philip Kobernik, Alexander Clegg, Paul Calamia, Dhruv Batra, Philip W Robin- son, and Kristen Grauman. SoundSpaces 2.0: A Simulation Platform for Visual-Acoustic Learning. In NeurIPS Datasets and Benchmarks Track, 2022. 2, 3

Show all 58 references
  1. [9]

    Novel-View Acoustic Synthesis

    Changan Chen, Alexander Richard, Roman Shapovalov, Vamsi Krishna Ithapu, Natalia Neverova, Kristen Grauman, and Andrea Vedaldi. Novel-View Acoustic Synthesis. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 6409–6419, 2023. 3

  2. [10]

    Sim2Real Transfer for Audio-Visual Navi- gation with Frequency-Adaptive Acoustic Field Prediction

    Chen, Changan and Ramos, Jordi and Tomar, Anshul and Grauman, Kristen. Sim2Real Transfer for Audio-Visual Navi- gation with Frequency-Adaptive Acoustic Field Prediction. In The IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2024. 3

  3. [11]

    Adverb: Visually guided audio dereverberation

    Sanjoy Chowdhury, Sreyan Ghosh, Subhrajyoti Dasgupta, An- ton Ratnarajah, Utkarsh Tyagi, and Dinesh Manocha. Adverb: Visually guided audio dereverberation. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 7884–7896, 2023. 2, 5, 18

  4. [12]

    Amen- gual Gari

    Orchisama Das, Paul Calamia, and Sebastia V . Amen- gual Gari. Room Impulse Response Interpolation from a Sparse Set of Measurements Using a Modal Architecture. In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 960–964, 2021. 2

  5. [13]

    Tenenbaum

    Chuang Gan, Yiwei Zhang, Jiajun Wu, Boqing Gong, and Joshua B. Tenenbaum. Look, Listen, and Act: Towards Audio- Visual Embodied Navigation. In IEEE International Confer- ence on Robotics and Automation (ICRA), pages 9701–9707,

  6. [14]

    Gumerov and Ramani Duraiswami

    Nail A. Gumerov and Ramani Duraiswami. A broadband fast multipole accelerated boundary element method for the three dimensional Helmholtz equation. The Journal of the Acoustical Society of America, 125(1):191–205, 2009. 2

  7. [15]

    FDTD Methods for 3-D Room Acoustics Simulation With High-Order Accuracy in Space and Time

    Brian Hamilton and Stefan Bilbao. FDTD Methods for 3-D Room Acoustics Simulation With High-Order Accuracy in Space and Time. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 25(11):2112–2124, 2017. 2

  8. [16]

    F.J. Harris. On the use of windows for harmonic analysis with the discrete Fourier transform. Proceedings of the IEEE, 66 (1):51–83, 1978. 5

  9. [17]

    Immersive Spatial Audio Reproduction for VR/AR Using Room Acoustic Modelling from 360 Images

    Hansung Kim, Luca Remaggi, Philip JB Jackson, and Adrian Hilton. Immersive Spatial Audio Reproduction for VR/AR Using Room Acoustic Modelling from 360 Images. In IEEE Conference on Virtual Reality and 3D User Interfaces (VR), pages 120–126. IEEE, 2019. 2

  10. [18]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. Adam: A Method for Stochastic Optimization. In 3rd International Conference on Learning Representations (ICLR), 2015. 5

  11. [19]

    Deep neural networks for cross-modal estimations of acoustic reverberation character- istics from two-dimensional images

    Homare Kon and Hideki Koike. Deep neural networks for cross-modal estimations of acoustic reverberation character- istics from two-dimensional images. In Audio Engineering Society Convention 144. Audio Engineering Society, 2018. 3

  12. [20]

    An auditory scaling method for reverb synthesis from a single two-dimensional image

    Homare Kon and Hideki Koike. An auditory scaling method for reverb synthesis from a single two-dimensional image. Acoustical Science and Technology, 41(4):675–685, 2020. 3

  13. [21]

    Virtual reality system with integrated sound field simulation and reproduction

    Tobias Lentz, Dirk Schr¨oder, Michael V orl¨ander, and Ingo As- senmacher. Virtual reality system with integrated sound field simulation and reproduction. EURASIP journal on advances in signal processing, 2007:1–19, 2007. 18

  14. [22]

    Scene-aware audio for 360 videos

    Dingzeyu Li, Timothy R Langlois, and Changxi Zheng. Scene-aware audio for 360 videos. ACM Transactions on Graphics (TOG), 37(4):1–12, 2018. 3

  15. [23]

    A V-NeRF: Learning Neural Fields for Real-World Audio-Visual Scene Synthesis

    Susan Liang, Chao Huang, Yapeng Tian, Anurag Kumar, and Chenliang Xu. A V-NeRF: Learning Neural Fields for Real-World Audio-Visual Scene Synthesis. In Conference on Neural Information Processing Systems (NeurIPS), 2023. 2

  16. [24]

    Neural Acoustic Context Field: Rendering Realistic Room Impulse Response With Neural Fields, 2023

    Susan Liang, Chao Huang, Yapeng Tian, Anurag Kumar, and Chenliang Xu. Neural Acoustic Context Field: Rendering Realistic Room Impulse Response With Neural Fields, 2023. 2, 3

  17. [25]

    Sound synthesis, propa- gation, and rendering

    Shiguang Liu and Dinesh Manocha. Sound synthesis, propa- gation, and rendering. Morgan & Claypool Publishers, 2022. 2

  18. [26]

    Tarr, Joshua B

    Andrew Luo, Yilun Du, Michael J. Tarr, Joshua B. Tenen- baum, Antonio Torralba, and Chuang Gan. Learning Neural Acoustic Fields. In Advances in Neural Information Process- ing Systems (NeurIPS), 2022. 2, 3

  19. [27]

    Few-Shot Audio-Visual Learning of En- vironment Acoustics

    Sagnik Majumder, Changan Chen*, Ziad Al-Halah*, and Kristen Grauman. Few-Shot Audio-Visual Learning of En- vironment Acoustics. In Conference on Neural Information Processing Systems (NeurIPS), 2022. 3, 5, 6, 13, 14, 17, 18

  20. [28]

    Lin, and Dinesh Manocha

    Ravish Mehra, Nikunj Raghuvanshi, Lauri Savioja, Ming C. Lin, and Dinesh Manocha. An efficient GPU-based time do- 9 main solver for the acoustic wave equation.Applied Acoustics, 73(2):83–94, 2012. 2

  21. [29]

    A V-SAM: Segment Anything Model Meets Audio-Visual Localization and Segmentation

    Shentong Mo and Yapeng Tian. A V-SAM: Segment Anything Model Meets Audio-Visual Localization and Segmentation. arXiv preprint arXiv:2305.01836, 2023. 2

  22. [30]

    Librispeech: an asr corpus based on public do- main audio books

    Vassil Panayotov, Guoguo Chen, Daniel Povey, and Sanjeev Khudanpur. Librispeech: an asr corpus based on public do- main audio books. In 2015 IEEE international conference on acoustics, speech and signal processing (ICASSP), pages 5206–5210. IEEE, 2015. 12

  23. [31]

    Nikunj Raghuvanshi, Rahul Narain, and Ming C. Lin. Effi- cient and Accurate Sound Propagation Using Adaptive Rect- angular Decomposition. IEEE Transactions on Visualization and Computer Graphics (TVCG), 15(5):789–801, 2009. 2

  24. [32]

    Listen2Scene: Inter- active material-aware binaural sound propagation for recon- structed 3D scenes

    Anton Ratnarajah and Dinesh Manocha. Listen2Scene: Inter- active material-aware binaural sound propagation for recon- structed 3D scenes . In IEEE Conference Virtual Reality and 3D User Interfaces (VR), pages 254–264, 2024. 2, 3

  25. [33]

    IR- GAN: Room Impulse Response Generator for Far-Field Speech Recognition

    Anton Ratnarajah, Zhenyu Tang, and Dinesh Manocha. IR- GAN: Room Impulse Response Generator for Far-Field Speech Recognition. In Proceedings of Interspeech 2021 , pages 286–290, 2021. 2

  26. [34]

    TS- RIR: Translated Synthetic Room Impulse Responses for Speech Augmentation

    Anton Ratnarajah, Zhenyu Tang, and Dinesh Manocha. TS- RIR: Translated Synthetic Room Impulse Responses for Speech Augmentation. In IEEE Automatic Speech Recog- nition and Understanding Workshop (ASRU), pages 259–266,

  27. [35]

    Fast-RIR: Fast Neural Dif- fuse Room Impulse Response Generator

    Anton Ratnarajah, Shi-Xiong Zhang, Meng Yu, Zhenyu Tang, Dinesh Manocha, and Dong Yu. Fast-RIR: Fast Neural Dif- fuse Room Impulse Response Generator. In IEEE Interna- tional Conference on Acoustics, Speech and Signal Process- ing (ICASSP), pages 571–575, 2022. 2, 3, 6, 13, 14, 17

  28. [36]

    Towards improved room impulse response estimation for speech recognition

    Anton Ratnarajah, Ishwarya Ananthabhotla, Vamsi Krishna Ithapu, Pablo Hoffmann, Dinesh Manocha, and Paul Calamia. Towards improved room impulse response estimation for speech recognition. In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pa...

  29. [37]

    A V-RIR: Audio-Visual Room Impulse Response Estimation

    Anton Ratnarajah, Sreyan Ghosh, Sonal Kumar, Purva Chiniya, and Dinesh Manocha. A V-RIR: Audio-Visual Room Impulse Response Estimation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 27164–27175, 2024. 2, 3, 6, 13, 14, 17, 18

  30. [38]

    Reproducing real world acoustics in virtual reality using spherical cameras

    Luca Remaggi, Hansung Kim, Philip JB Jackson, and Adrian Hilton. Reproducing real world acoustics in virtual reality using spherical cameras. In International Conference on Immersive and Interactive Audio. Audio Engineering Society,

  31. [39]

    V oices Obscured in Complex Environmental Settings (VOICES) corpus

    Colleen Richey, Maria A Barrios, Zeb Armstrong, Chris Bartels, Horacio Franco, Martin Graciarena, Aaron Lawson, Mahesh Kumar Nandwana, Allen Stauffer, Julien van Hout, et al. V oices Obscured in Complex Environmental Settings (VOICES) corpus. arXiv preprint arXiv:1804.05053, 2018. 12

  32. [40]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer-Assisted Inter- vention (MICCAI), pages 234–241. Springer, 2015. 4, 16

  33. [41]

    Interactive Sound Prop- agation and Rendering for Large Multi-Source Scenes

    Carl Schissler and Dinesh Manocha. Interactive Sound Prop- agation and Rendering for Large Multi-Source Scenes. ACM Transactions on Graphics (TOG), 36(4):1, 2016. 2, 3

  34. [42]

    Acous- tic classification and optimization for multi-modal rendering of real-world scenes

    Carl Schissler, Christian Loftin, and Dinesh Manocha. Acous- tic classification and optimization for multi-modal rendering of real-world scenes. IEEE Transactions on Visualization and Computer Graphics (TVCG), 24(3):1246–1259, 2017. 3

  35. [43]

    Acous- tic Classification and Optimization for Multi-Modal Render- ing of Real-World Scenes

    Carl Schissler, Christian Loftin, and Dinesh Manocha. Acous- tic Classification and Optimization for Multi-Modal Render- ing of Real-World Scenes. IEEE Transactions on Visual- ization and Computer Graphics (TVCG), 24(3):1246–1259,

  36. [44]

    Image2Reverb: Cross-Modal Reverb Impulse Response Synthesis

    Nikhil Singh, Jeff Mentch, Jerry Ng, Matthew Beveridge, and Iddo Drori. Image2Reverb: Cross-Modal Reverb Impulse Response Synthesis. IEEE/CVF International Conference on Computer Vision (ICCV), pages 286–295, 2021. 2, 3, 6, 13, 14, 17

  37. [45]

    Self-Supervised Visual Acoustic Matching

    Arjun Somayazulu, Changan Chen, and Kristen Grauman. Self-Supervised Visual Acoustic Matching. Advances in Neu- ral Information Processing Systems (NeurIPS), 36, 2024. 3

  38. [46]

    Dropout: a simple way to prevent neural networks from overfitting

    Nitish Srivastava, Geoffrey Hinton, Alex Krizhevsky, Ilya Sutskever, and Ruslan Salakhutdinov. Dropout: a simple way to prevent neural networks from overfitting. The journal of machine learning research, 15(1):1929–1958, 2014. 16

  39. [47]

    INRAS: Implicit Neural Representation for Audio Scenes

    Kun Su, Mingfei Chen, and Eli Shlizerman. INRAS: Implicit Neural Representation for Audio Scenes. In Advances in Neural Information Processing Systems (NeurIPS), 2022. 2, 3

  40. [48]

    Habitat 2.0: Training Home Assistants to Rearrange their Habitat

    Andrew Szot, Alex Clegg, Eric Undersander, Erik Wijmans, Yili Zhao, John Turner, Noah Maestre, Mustafa Mukadam, Devendra Chaplot, Oleksandr Maksymets, Aaron Gokaslan, Vladimir V ondrus, Sameer Dharur, Franziska Meier, Wojciech Galuba, Angel Chang, Zsolt Kira, Vladlen Koltun, J...

  41. [49]

    Bryan, Dingzeyu Li, Timothy R

    Zhenyu Tang, Nicholas J. Bryan, Dingzeyu Li, Timothy R. Langlois, and Dinesh Manocha. Scene-aware audio rendering via deep acoustic analysis. IEEE Transactions on Visualiza- tion and Computer Graphics (TVCG), 26:1991–2001, 2019. 2

  42. [50]

    GW A: A large high-quality acoustic dataset for audio processing

    Zhenyu Tang, Rohith Aralikatti, Anton Jeran Ratnarajah, and Dinesh Manocha. GW A: A large high-quality acoustic dataset for audio processing. InProceedings of the ACM Special Inter- est Group on Computer Graphics and Interactive Techniques (SIGGRAPH), pages 1–9, 2022. 3

  43. [51]

    Thompson

    Lonny L. Thompson. A review of finite-element methods for time-harmonic acoustics. The Journal of the Acoustical Society of America, 119(3):1315–1330, 2006. 2

  44. [52]

    Audio-Visual Event Localization in Unconstrained Videos

    Yapeng Tian, Jing Shi, Bochen Li, Zhiyao Duan, and Chen- liang Xu. Audio-Visual Event Localization in Unconstrained Videos. In Proceedings of the European Conference on Com- puter Vision (ECCV), 2018. 2

  45. [53]

    Building acoustics

    Tor Erik Vigran. Building acoustics. CRC Press, 2014. 1, 18 10

  46. [54]

    Simulation of the transient and steady- state sound propagation in rooms using a new combined ray- tracing/image-source algorithm

    Michael V orl¨ander. Simulation of the transient and steady- state sound propagation in rooms using a new combined ray- tracing/image-source algorithm. The Journal of the Acoustical Society of America, 86(1):172–178, 1989. 2

  47. [55]

    Creation of auditory augmented reality using a position- dynamic binaural synthesis system—technical components, psychoacoustic needs, and perceptual evaluation

    Stephan Werner, Florian Klein, Annika Neidhardt, Ulrike Sloma, Christian Schneiderwind, and Karlheinz Branden- burg. Creation of auditory augmented reality using a position- dynamic binaural synthesis system—technical components, psychoacoustic needs, and perceptual evaluation...

  48. [56]

    Binau- ral Audio-Visual Localization

    Xinyi Wu, Zhenyao Wu, Lili Ju, and Song Wang. Binau- ral Audio-Visual Localization. In Proceedings of the AAAI Conference on Artificial Intelligence, pages 2961–2968, 2021. 2

  49. [57]

    Empirical Evaluation of Rectified Activations in Convolutional Network

    Bing Xu. Empirical Evaluation of Rectified Activations in Convolutional Network. arXiv preprint arXiv:1505.00853,

  50. [58]

    Bevformer v2: Adapting modern image backbones to bird’s-eye-view recognition via perspective su- pervision

    Chenyu Yang, Yuntao Chen, Hao Tian, Chenxin Tao, Xizhou Zhu, Zhaoxiang Zhang, Gao Huang, Hongyang Li, Yu Qiao, Lewei Lu, et al. Bevformer v2: Adapting modern image backbones to bird’s-eye-view recognition via perspective su- pervision. In Proceedings of the IEEE/CVF conference...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.