A Cross-Evaluated Database of Measured and Simulated HRTFs Including 3D Head Meshes, Anthropometric Features, and Headphone Impulse Responses

Brinkmann, Fabian; Dinakaran, Manoj; Pelzer, Robert; Grosche, Peter; Voss, Daniel; Weinzierl, Stefan

doi:10.17743/jaes.2019.0024

Cited by 67 publications

(60 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A first non-public version of the database was created in 2016 to provide input data to the participants of a round robin investigation on room acoustic simulation and auralization [32]. At this stage, the database did not include the measured impulse responses.…”

Section: Warningmentioning

confidence: 99%

A benchmark for room acoustical simulation. Concept and database

et al. 2021

Self Cite

View full text Add to dashboard Cite

Section: Warningmentioning

confidence: 99%

A benchmark for room acoustical simulation. Concept and database

et al. 2021

Self Cite

View full text Add to dashboard Cite

“…We use the HUTUBS dataset [24] in this work as it is one of the few that contains all the required assets. The dataset contains 96 subjects' measured head-related impulse responses (HRIRs), 58 head meshes, and a set of 25 anthropometric parameters for each subject.…”

Section: Datasetmentioning

confidence: 99%

“…The 3D head models are adaptively remeshed to ensure solver performance. The procedure is identical to that described in [24,28], which gradually coarsens the mesh from the ipsilateral ear at 1 mm to the contralateral ear at 10 mm. We were concerned that the resolution on the 10 mm side might be too coarse, so we perform studies on meshes at 1 mm-8 mm, 1 mm-6 mm, 1 mm-5 mm, and 1 mm-4 mm resolutions.…”

Section: Numerical Simulationmentioning

confidence: 99%

“…This can be done, for example, by feeding a subject's anthropometric parameters through neural networks [17][18][19][20][21] to get HRTF predictions. However, the quality of reconstruction depends on the accuracy and size of the underlying dataset, which is often quite small [8,[22][23][24][25] compared to other machine learning applications. This limits how expressive the models can be and consequently the ability to model large variations in personalized HRTFs.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Personalized HRTF Modeling Using DNN-Augmented BEM

Zhang

Wang

James

2021

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

Accurate modeling of personalized head-related transfer functions (HRTFs) is difficult but critical for applications requiring spatial audio. However, this remains challenging as experimental measurements require specialized equipment, numerical simulations require accurate head geometries and robust solvers, and data-driven methods are hungry for data. In this paper, we propose a new deep learning method that combines measurements and numerical simulations to take the best of three worlds. By learning the residual difference and establishing a high quality spatial basis, our method achieves consistently 2 dB to 2.5 dB lower spectral distortion (SD) compared to the state-of-the-art methods.

show abstract

“…These are susceptible to domain discrepancy between synthesized ear shapes and actual human ear shapes (because the span of simulated ear shapes might be different from the span/manifold of HRTFs). Another recent open-sourced dataset of simulated HRTFs (from 3D scan of subjects) [19] also has small size (96 subjects).…”

Section: Introductionmentioning

confidence: 99%

On the Predictability of Hrtfs from Ear Shapes Using Deep Networks

Zhou

Jiang

Ithapu

2021

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

Head-Related Transfer Function (HRTF) individualization is critical for immersive and realistic spatial audio rendering in augmented/virtual reality. Neither measurements nor simulations using 3D scans of head/ear are scalable for practical applications. More efficient machine learning approaches are being explored recently, to predict HRTFs from ear images or anthropometric features. However, it is not yet clear whether such models can provide an alternative for direct measurements or high-fidelity simulations. Here, we aim to address this question. Using 3D ear shapes as inputs, we explore the bounds of HRTF predictability using deep neural networks. To that end, we propose and evaluate two models, and identify the lowest achievable spectral distance error when predicting the true HRTF magnitude spectra.

show abstract

A Cross-Evaluated Database of Measured and Simulated HRTFs Including 3D Head Meshes, Anthropometric Features, and Headphone Impulse Responses

Abstract: 2019). "A Cross-validated database of measured and simulated HRTFs including 3D head meshes and anthropometric features." J. Audio Eng. Soc.

Cited by 67 publications

References 0 publications

A benchmark for room acoustical simulation. Concept and database

A benchmark for room acoustical simulation. Concept and database

Personalized HRTF Modeling Using DNN-Augmented BEM

On the Predictability of Hrtfs from Ear Shapes Using Deep Networks

Contact Info

Product

Resources

About