GWA: A Large High-Quality Acoustic Dataset for Audio Processing

Tang, Zhenyu; Aralikatti, Rohith; Ratnarajah, Anton; Manocha, Dinesh

doi:10.48550/arxiv.2204.01787

Cited by 3 publications

(14 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The publicly available IR datasets are either recorded in a realworld environments (recorded IR) [Carlo et al 2021;Hadad et al 2014;Koyama et al 2021; or generated using IR simulation tools (synthetic IR) Grondin et al 2020;Ko et al 2017;Tang et al 2022]. The recorded IR datasets are limited in size (fewer than 5000 IRs) and number of environments (fewer than 10 rooms), and no sufficient information on the recording conditions is provided to train a deep learning model.…”

Section: Indoor 3d Scenes and Ir Databasesmentioning

confidence: 99%

“…The recorded IR datasets are limited in size (fewer than 5000 IRs) and number of environments (fewer than 10 rooms), and no sufficient information on the recording conditions is provided to train a deep learning model. Synthetic IR datasets like SoundSpaces and GWA [Tang et al 2022] contain millions of IRs. The IRs in SoundSpaces is simulated using a geometric acoustic method [Schissler and Manocha 2011] and GWA dataset contains high-quality IRs computed using a hybrid method.…”

Section: Indoor 3d Scenes and Ir Databasesmentioning

confidence: 99%

“…In our work, we train MESH2IR on the GWA IR dataset [Tang et al 2022] simulated using a hybrid algorithm to generate realistic IRs on unseen complex environments. The IRs in GWA are created by automatically calibrating the ray energies simulated using the geometric acoustic method [Tang et al 2020] with the wave effects simulated using finite-difference time-domain wave solver [Botteldooren 1995] to create high-quality low-frequency and high-frequency wave effects.…”

Section: Training Datasetmentioning

confidence: 99%

“…At low frequencies under 500 Hz, the ray assumption is not valid for most scenes and results in significant simulation errors. Hybrid sound propagation algorithms [Tang et al 2022;Yeh et al 2013] combine IRs from wave-based and geometric techniques to generate accurate IRs in the human aural range for complex dynamic scenes. Generating IRs corresponding to thousands of sound sources at an interactive rate is not possible with the hybrid method because of its complexity.…”

Section: Introductionmentioning

confidence: 99%

“…We also evaluate the contribution of energy decay relief in improving the IRs generated from MESH2IR. We train our MESH2IR on an IR dataset computed using a hybrid sound propagation algorithm [Tang et al 2022]. Therefore, the predicted IRs from MESH2IR exhibit good accuracy for both low-frequency and high-frequency components.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

MESH2IR: Neural Acoustic Impulse Response Generator for Complex 3D Scenes

Ratnarajah,

Tang,

Aralikatti

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

We propose a mesh-based neural network (MESH2IR) to generate acoustic impulse responses (IRs) for indoor 3D scenes represented using a mesh. The IRs are used to create a high-quality sound experience in interactive applications and audio processing. Our method can handle input triangular meshes with arbitrary topologies (2K -3M triangles). We present a novel training technique to train MESH2IR using energy decay relief and highlight its benefits. We also show that training MESH2IR on IRs preprocessed using our proposed technique significantly improves the accuracy of IR generation. We reduce the non-linearity in the mesh space by transforming 3D scene meshes to latent space using a graph convolution network. Our MESH2IR is more than 200 times faster than a geometric acoustic algorithm on a CPU and can generate more than 10,000 IRs per second on an NVIDIA GeForce RTX 2080 Ti GPU for a given furnished indoor 3D scene. The acoustic metrics are used to characterize the acoustic environment. We show that the acoustic metrics of the IRs predicted from our MESH2IR match the ground truth with less than 10% error. We also highlight the benefits of MESH2IR on audio and speech processing applications such as speech dereverberation and speech separation. To the best of our knowledge, ours is the first neural-network-based approach to predict IRs from a given 3D scene mesh in real-time.

show abstract

Section: Indoor 3d Scenes and Ir Databasesmentioning

confidence: 99%

Section: Indoor 3d Scenes and Ir Databasesmentioning

confidence: 99%

Section: Training Datasetmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

MESH2IR: Neural Acoustic Impulse Response Generator for Complex 3D Scenes

Ratnarajah,

Tang,

Aralikatti

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Realistic Sources, Receivers and Walls Improve The Generalisability of Virtually-Supervised Blind Acoustic Parameter Estimators

Srivastava

Deleforge

2022

2022 International Workshop on Acoustic Signal Enhancement (IWAENC)

View full text Add to dashboard Cite

Blind acoustic parameter estimation consists in inferring the acoustic properties of an environment from recordings of unknown sound sources. Recent works in this area have utilized deep neural networks trained either partially or exclusively on simulated data, due to the limited availability of real annotated measurements. In this paper, we study whether a model purely trained using a fast image-source room impulse response simulator can generalize to real data. We present an ablation study on carefully crafted simulated training sets that account for different levels of realism in source, receiver and wall responses. The extent of realism is controlled by the sampling of wall absorption coefficients and by applying measured directivity patterns to microphones and sources. A state-of-the-art model trained on these datasets is evaluated on the task of jointly estimating the room's volume, total surface area, and octave-band reverberation times from multiple, multichannel speech recordings. Results reveal that every added layer of simulation realism at train time significantly improves the estimation of all quantities on real signals.

show abstract

NeuralSound

Jin

Wang

et al. 2022

ACM Trans. Graph.

View full text Add to dashboard Cite

We present a novel learning-based modal sound synthesis approach that includes a mixed vibration solver for modal analysis and a radiation network for acoustic transfer. Our mixed vibration solver consists of a 3D sparse convolution network and a Locally Optimal Block Preconditioned Conjugate Gradient (LOBPCG) module for iterative optimization. Moreover, we highlight the correlation between a standard numerical vibration solver and our network architecture. Our radiation network predicts the Far-Field Acoustic Transfer maps (FFAT Maps) from the surface vibration of the object. The overall running time of our learning-based approach for most new objects is less than one second on a RTX 3080 Ti GPU while maintaining a high sound quality close to the ground truth solved by standard numerical methods. We also evaluate the numerical and perceptual accuracy of our approach on different objects with various shapes and materials.

show abstract

GWA: A Large High-Quality Acoustic Dataset for Audio Processing

Cited by 3 publications

References 38 publications

MESH2IR: Neural Acoustic Impulse Response Generator for Complex 3D Scenes

MESH2IR: Neural Acoustic Impulse Response Generator for Complex 3D Scenes

Realistic Sources, Receivers and Walls Improve The Generalisability of Virtually-Supervised Blind Acoustic Parameter Estimators

NeuralSound

Contact Info

Product

Resources

About