Room Reverberation Reconstruction: Interpolation of the Early Part Using Compressed Sensing

Mignot, Rémi; Daudet, Laurent; Ollivier, François

doi:10.1109/tasl.2013.2273662

Cited by 54 publications

(31 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In particular, Figs. 12(c), (g), and (k) also shows that the interpolation (using shearlets) is accurate in the late part of the responses, which on its own is an improvement with respect to previous studies [21,22]. As in the case of Fig.…”

Section: Linearsupporting

confidence: 55%

“…The benefit this framework has brought to the acoustics community is a reduction of microphones in various data acquisition and signal processing tasks [20], one of them being the task of interpolating RIRs. One of the pioneering papers was written by Mignot et al [21], in which a dictionary of virtual monopole sources is used to interpolate the early part of the RIRs -thereby exploiting the temporal sparsity of the early part, assuming the room is empty and there is no diffraction phenomena. A few years later, Antonello et al investigated the use of a dictionary of time-dependent equivalent sources in various sparsity-promoting domains [22], assuming far-field measurement distances, and performed an experimental validation in an empty, rectangular room.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Compressed sensing of impulse responses in rooms of unknown properties and contents

Zea

2019

Journal of Sound and Vibration

View full text Add to dashboard Cite

Section: Linearsupporting

confidence: 55%

Section: Introductionmentioning

confidence: 99%

Compressed sensing of impulse responses in rooms of unknown properties and contents

Zea

2019

Journal of Sound and Vibration

View full text Add to dashboard Cite

“…Aging and/or environmental effects such as temperature changes introduce system errors making calibration capability an unavoidable need for a given system. Furthermore, imperfect knowledge of the system parameters may cause calibration errors [6][7][8]. Guided calibration solutions can be used to achieve desired performance in which a known training signal is used.…”

Section: Introductionmentioning

confidence: 99%

Sparsity based off-grid blind sensor calibration

Çamlıca

Yetik

Arıkan

2019

Digital Signal Processing

View full text Add to dashboard Cite

Compressive Sensing (CS) based techniques generally discretize the signal space and assume that the signal has a sparse support restricted on the discretized grid points. This restriction of representing the signal on a discretized grid results in the off-grid problem which causes performance degradation in the reconstruction of signals. Sensor calibration is another issue which can cause performance degradation if not properly addressed. Calibration aims to reduce the disruptive effects of the phase and the gain biases. In this paper, a CS based blind calibration technique is proposed for the reconstruction of multiple off-grid signals. The proposed technique is capable of estimating the off-grid signals and correcting the gain and the phase biases due to insufficient calibration simultaneously. It is applied to off-grid frequency estimation and direction finding applications using blind calibration. Extensive simulation analyses are performed for both applications. Results show that the proposed technique has superior reconstruction performance.

show abstract

“…In such environments, the problem can be reformulated as that of predicting the RTF for a given pose given a set of RTFs or ATFs recorded for other poses. Recent studies have performed ATF interpolation based on room geometry estimation [13], compressed sensing [14,15] or models derived from the wave propagation equation [16]. * We thank CSC (201604910623) for funding.…”

Section: Introductionmentioning

confidence: 99%

Semi-Supervised Learning with Deep Neural Networks for Relative Transfer Function Inverse Regression

Wang

Yan

2018

2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

supervised learning with deep neural networks for relative transfer function inverse regression. ABSTRACTPrior knowledge of the relative transfer function (RTF) is useful in many applications but remains little studied. In this paper, we propose a semi-supervised learning algorithm based on deep neural networks (DNNs) for RTF inverse regression, that is to generate the full-band RTF vector directly from the source-receiver pose (position and orientation). Two typical scenarios are discussed: training on labeled RTFs only, or on additional unlabeled RTFs. Both setups utilize the low-dimensional manifold property of RTF in stationary environments. With this property as an additional regularization term, a smooth mapping solution with respect to the manifold is obtained. Experimental simulations show that the proposed method achieves a lower mean prediction error than the free field model with few labeled RTFs, and the unlabeled RTFs are essential in improving the inverse regression performance.Index Termsrelative transfer function, semi-supervised learning, deep neural network, manifold regularization

show abstract

Room Reverberation Reconstruction: Interpolation of the Early Part Using Compressed Sensing

Abstract: International audienc

Cited by 54 publications

References 21 publications

Compressed sensing of impulse responses in rooms of unknown properties and contents

Compressed sensing of impulse responses in rooms of unknown properties and contents

Sparsity based off-grid blind sensor calibration

Semi-Supervised Learning with Deep Neural Networks for Relative Transfer Function Inverse Regression

Contact Info

Product

Resources

About