In the context of loudspeaker design and manufacturing, impulse response precompensation is a broadly studied topic. While standard methods based on DSP techniques achieve satisfactory results, they require a tractable mathematical representation of the system at hand, often obtained through system identificationlike procedures. This thesis project investigates the possibility of approaching the same task from a different perspective, i.e. by employing neural networks, under the assumption that no analytical model of the loudspeaker is known. Neural architectures may be good candidates for tackling the problem given their strong representational power which makes them suitable especially in highly nonlinear spaces. In the context of deep learning, training an endtoend system typically requires information on the derivatives of all the functions involved. This work explores two different methods to accomplish impulse response precompensation using neural networks without gradients: the first one consists in efficiently computing an unbiased (in the limit) estimator of the gradient of the loudspeaker model using the Simultaneous Perturbation Stochastic Approximation (SPSA) algorithm, while the other consists in employing a second network to learn a differentiable approximation of the speaker function and using that to train the target precompensating network. Both the approaches will be compared to the baseline and the quality of the results are assessed through objective metrics directly measured on the output waveforms.
In the context of loudspeaker design and manufacturing, impulse response precompensation is a broadly studied topic. While standard methods based on DSP techniques achieve satisfactory results, they require a tractable mathematical representation of the system at hand, often obtained through system identificationlike procedures. This thesis project investigates the possibility of approaching the same task from a different perspective, i.e. by employing neural networks, under the assumption that no analytical model of the loudspeaker is known. Neural architectures may be good candidates for tackling the problem given their strong representational power which makes them suitable especially in highly nonlinear spaces. In the context of deep learning, training an endtoend system typically requires information on the derivatives of all the functions involved. This work explores two different methods to accomplish impulse response precompensation using neural networks without gradients: the first one consists in efficiently computing an unbiased (in the limit) estimator of the gradient of the loudspeaker model using the Simultaneous Perturbation Stochastic Approximation (SPSA) algorithm, while the other consists in employing a second network to learn a differentiable approximation of the speaker function and using that to train the target precompensating network. Both the approaches will be compared to the baseline and the quality of the results are assessed through objective metrics directly measured on the output waveforms.
Deep learningbased loudspeaker compensation system using gradientfree stochastic optimization
TRABUCCO, GIOVANNI
2022/2023
Abstract
In the context of loudspeaker design and manufacturing, impulse response precompensation is a broadly studied topic. While standard methods based on DSP techniques achieve satisfactory results, they require a tractable mathematical representation of the system at hand, often obtained through system identificationlike procedures. This thesis project investigates the possibility of approaching the same task from a different perspective, i.e. by employing neural networks, under the assumption that no analytical model of the loudspeaker is known. Neural architectures may be good candidates for tackling the problem given their strong representational power which makes them suitable especially in highly nonlinear spaces. In the context of deep learning, training an endtoend system typically requires information on the derivatives of all the functions involved. This work explores two different methods to accomplish impulse response precompensation using neural networks without gradients: the first one consists in efficiently computing an unbiased (in the limit) estimator of the gradient of the loudspeaker model using the Simultaneous Perturbation Stochastic Approximation (SPSA) algorithm, while the other consists in employing a second network to learn a differentiable approximation of the speaker function and using that to train the target precompensating network. Both the approaches will be compared to the baseline and the quality of the results are assessed through objective metrics directly measured on the output waveforms.File  Dimensione  Formato  

Trabucco_Giovanni.pdf
accesso riservato
Dimensione
8.87 MB
Formato
Adobe PDF

8.87 MB  Adobe PDF 
The text of this website © Università degli studi di Padova. Full Text are published under a nonexclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/46081