Skip to main content

SMART SOUND PROCESSING FOR THE DIGITAL LIVING

TEC2015-67387-C4-1-R

•
Funding agency name Ministerio de Economía y Competitividad
Funding agency acronym MINECO
Program Programa Estatal de I+D+i Orientado a los Retos de la Sociedad
Call Proyectos I+D+i Retos de Investigación
Call year 2015
State Plan 2013-2016 Plan Estatal de Investigación Científica y Técnica y de Innovación
Management unit Dirección General de Investigación Científica y Técnica
Beneficiary centre UNIVERSITAT POLITÈCNICA DE VALÈNCIA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 25
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Parallel Online Time Warping for Real-Time Audio-to-Score Alignment in Multi-core Systems

  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • Cortina, Raquel
  • Rodríguez-Serrano, F.J.
  • Vera-Candeas, P.
  • Alonso-González, M.
  • Ranilla, José
[EN] The Audio-to-Score framework consists of two separate stages: pre- processing and alignment. The alignment is commonly solved through offline Dynamic Time Warping (DTW), which is a method to find the path over the distortion matrix with the minimum cost to determine the relation between the performance and the musical score times. In this work we propose a par- allel online DTW solution based on a client-server architecture. The current version of the application has been implemented for multi-core architectures (x86, x64 and ARM), thus covering either powerful systems or mobile devices. An extensive experimentation has been conducted in order to validate the software. The experiments also show that our framework allows to achieve a good score alignment within the real-time window by using parallel computing techniques., This work has been partially supported by Spanish Ministry of Science and Innovation and FEDER under Projects TEC2012-38142-C04-01, TEC2012-38142-C04-03, TEC2012-38142-C04-04, TEC2015-67387-C4-1-R, TEC2015-67387-C4-3-R, TEC2015-67387-C4-4-R, the European Union FEDER (CAPAP-H5 network TIN2014-53522-REDT), and the Generalitat Valenciana under Grant PROMETEOII/2014/003.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

HReMAS: Hybrid Real-time Musical Alignment System

  • Cabañas-Molero, P.
  • Cortina-Parajón, Raquel
  • Combarro, E. F.
  • Bris-Peñalver, F. J.
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
[EN] This paper presents a real-time audio-to-score alignment system for musical applications. The aim of these systems is to synchronize a live musical performance with its symbolic representation in a music sheet. We have used as a base our previous real-time alignment system by enhancing it with a traceback stage, a stage used in offline alignment to improve the accuracy of the aligned note. This stage introduces some delay, what forces to assume a trade-off between output delay and alignment accuracy that must be considered in the design of this type of hybrid techniques. We have also improved our former system to execute faster in order to minimize this delay. Other interesting improvements, like identification of silence frames, have also been incorporated to our proposed system., This work has been supported by the "Ministerio de Economia y Competitividad" of Spain and FEDER under Projects TEC2015-67387-C4-{1,2,3}-R.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

An efficient musical accompaniment parallel system for mobile devices

  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • Vera-Candeas, P.
  • Cortina, Raquel
  • Ranilla, José
[EN] This work presents a software system designed to track the reproduction of a musical piece with the aim to match the score position into its symbolic representation on a digital sheet. Into this system, known as automated musical accompaniment system, the process of score alignment can be carried out real-time. A real-time score alignment, also known as score following, poses an important challenge due to the large amount of computation needed to process each digital frame and the very small time slot to process it. Moreover, the challenge is even greater since we are interested on handheld devices, i.e. devices characterized by both low power consumption and mobility. The results presented here show that it is possible to exploit efficiently several cores of an ARM(A (R)) processor, or a GPU accelerator (presented in some SoCs from NVIDIA) reducing the processing time per frame under 10 ms in most of the cases., This work was supported by the Ministry of Economy and Competitiveness from Spain (FEDER) under projects TEC2015-67387-C4-1-R, TEC2015-67387-C4-2-R and TEC2015-67387-C4-3-R, the Andalusian Business, Science and Innovation Council under project P2010-TIC-6762 (FEDER), and the Generalitat Valenciana PROMETEOII/2014/003
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

High-performance computing: the essential tool and the essential challenge

  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • Ranilla, José
  • Vigo-Aguiar, Jesús
[EN] Prolog to the Journal of Supercomputing, volume 73, issue 1., We would also like to acknowledge to the “Ministerio de Educación y Ciencia” of Spain, for its support to the Spanish CAPAP-H5 network (HPC in Heterogeneous Systems, TIN2014-53522-REDT), and to the “Ministerio de Economía y Competitividad” from Spain/FEDER for supporting Grants TEC2015-67387-C4-1-R and TEC2015-67387-C4-3-R.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Shaping zones of quiet in a large enclosure generated by an active noise control system

  • Wrona, Stanislaw
  • Pawelczyk, Marek
  • de Diego Antón, María|||0000-0001-9948-3396
[EN] Performance of an Active Noise Control (ANC) system strongly depends on sensors and actuators spatial arrangement. It determines both achieved Noise Reduction (NR) levels and spatial distribution of obtained zones of quiet, making it an essential problem. However, if the acoustic field in the enclosure can be appropriately modelled, then optimization algorithms can be employed to find efficient configuration of the ANC system, enhancing its performance according to a formulated cost function and constraints. This paper proposes a complete method for enhancing NR levels and shaping zones of quiet generated with an ANC system by optimization of sensors and actuators arrangement. A Memetic Algorithm (MA) is utilized. The MA itself and its proposed operators are described. The optimization problem formulation is derived and discussed. As a control algorithm, Distributed Multiple Error Filtered-x Least Mean Square (DMEFxLMS) is used. Extensive simulation results are presented for an exemplary real enclosure. The model of the acoustic environment has been obtained by real-world experiments, resulting in identification of 36864 acoustic responses in total. Practically feasible cost function and constraints are evaluated. Advantages and limits of the method are pointed out and discussed., The research reported in this paper has been supported by the National Science Centre, Poland, decision no. DEC-2017/25/B/ST7/02236, and by EU together with Spanish Government under Grant TEC2015-67387-C4-1-R b(MINECO/FEDER).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

A Pipeline for the QR Update in Digital Signal Processing

  • Dolz, Manuel F.
  • Alventosa, Fran J.
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • Vidal Maciá, Antonio Manuel
[EN] The input and output signals of a digital signal processing system can often be represented by a rectangular matrix as it is the case of the beamformer algorithm, a very useful particular algorithm that allows extraction of the original input signal once it is cleaned from noise and room reverberation. We use a version of this algorithm in which the system matrix must be factorized to solve a least squares problem. The matrix changes periodically according to the input signal sampled; therefore, the factorization needs to be recalculated as fast as possible. In this paper, we propose to use parallelism through a pipeline pattern. With our pipeline, some partial computations are advanced so that the final time required to update the factorization is highly reduced, This work was supported by the Spanish Ministry of Economy and Competitiveness under MINECO and FEDER projects TIN2014-53495-R and TEC2015-67387-C4-1-R.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Solving Weighted Least Squares (WLS) problems on ARM-based architectures

  • Belloch Rodríguez, José Antonio
  • Bank, Balazs
  • Igual Peña, Francisco Daniel
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
  • Vidal Maciá, Antonio Manuel
TheWeighted Least Squares algorithm (WLS) is applied to numerous optimization problems, but requires the use of high computational resources, especially when complex arithmetic is involved. This work aims to accelerate the resolution of a WLS problem by reducing the computational cost (relaying on BLAS/LAPACK routines) and the computational precision from double to single. As a test case, we design an IIR filter for a Graphic Equalizer, where the numerical errors due to single precision are easily visualized. In addition, given the importance of low power architectures for this kind of implementations, we evaluate the performance, scalability, and energy efficiency of each method on two different processors implementing the ARMv7 architecture, widely used in current mobile devices with power constraints. Results show that the method that exhibits a high theoretical computational cost overcomes in efficiency other methods with lower theoretical cost in architectures of this type., This work started in spring 2016 when Jose A. Belloch was a visiting postdoctoral researcher at Budapest University of Technology and Economics thanks to the European Network COST Action IC1305 inside the program Short Term Scientific Mission with the following reference: COST-SPASM-ECOST-STSM-IC1305-020416-072431. Dr. Jose A. Belloch is supported by GVA contract APOSTD/2016/069. The researchers from Universitat Jaume I are supported by the CICYT projects TIN2014-53495-R of MINECO and FEDER. The authors from the Universitat Politecnica de Valencia are supported by MINECO Projects TEC2015-67387-C4-1-R, PROMETEOII/2014/003 and CAPAP-H5 network TIN2014-53522-REDT. The researcher from UCM is supported by the EU (FEDER) and the Spanish MINECO, under Grants TIN 2015-65277-R and TIN2012-32180. The work of Balazs Bank was supported by the UNKP-16-4-III New National Excellence Program of the Ministry of Human Capacities, Hungary.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

GPU-Based Dynamic Wave Field Synthesis Using Fractional Delay Filters and Room Compensation

  • Belloch, Jose A.
  • Quintana-Orti, Enrique S.
  • Välimäki, Vesa
  • Gonzalez, Alberto|||0000-0002-6984-3212
  • Ferrer Contreras, Miguel|||0000-0002-8743-1887
[EN] Wave field synthesis (WFS) is a multichannel audio reproduction method, of a considerable computational cost that renders an accurate spatial sound field using a large number of loudspeakers to emulate virtual sound sources. The moving of sound source locations can be improved by using fractional delay filters, and room reflections can be compensated by using an inverse filter bank that corrects the room effects at selected points within the listening area. However, both the fractional delay filters and the room compensation filters further increase the computational requirements of the WFS system. This paper analyzes the performance of a WFS system composed of 96 loudspeakers which integrates both strategies. In order to deal with the large computational complexity, we explore the use of a graphics processing unit (GPU) as a massive signal co-processor to increase the capabilities of the WFS system. The performance of the method as well as the benefits of the GPU acceleration are demonstrated by considering different sizes of room compensation filters and fractional delay filters of order 9. The results show that a 96-speaker WFS system that is efficiently implemented on a state-of-art GPU can synthesize the movements of 94 sound sources in real time and, at the same time, can manage 9216 room compensation filters having more than 4000 coefficients each., This work was supported in part by the European Union together with Spanish Government under Grants TEC2015-67387-C41-R and TIN2014-53495-R (MINECO/FEDER) and in part by the Generalitat Valenciana under Grant PROMETEOII/2014/003. The work of J. A. Belloch was supported by the postdoctoral fellowship from Generalitat Valenciana under Grant APOSTD/2016/069
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Blockwise Frequency Domain Active Noise Controller Over Distributed Networks

  • Antoñanzas-Manuel, Christian
  • Ferrer Contreras, Miguel|||0000-0002-8743-1887
  • de Diego Antón, María|||0000-0001-9948-3396
  • Gonzalez, Alberto|||0000-0002-6984-3212
This work presents a practical active noise control system composed of distributed and collaborative acoustic nodes. To this end, experimental tests have been carried out in a listening room with acoustic nodes equipped with loudspeakers and microphones. The communication among the nodes is simulated by software. We have considered a distributed algorithm based on the Filtered-x Least Mean Square (FxLMS) method that introduces collaboration between nodes following an incremental strategy. For improving the processing efficiency in practical scenarios where data acquisition systems work by blocks of samples, the frequency-domain partitioned block technique has been used. Implementation aspects such as computational complexity, processing time of the network and convergence of the algorithm have been analyzed. Experimental results show that, without constraints in the network communications, the proposed distributed algorithm achieves the same performance as the centralized version. The performance of the proposed algorithm over a network with a given communication delay is also included., This work has been supported by the European Union (European Regional Development Fund) together with Spanish Government through TEC2015-67387-C4-1-R project, the grant BES-2013-063783 and Generalitat Valenciana through the PROMETEOII/2014/003 project.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Exploring Hybrid Parallel Systems for Probabilistic Record Linkage

  • Boratto, Murilo
  • Pinto, Clicia
  • Melo, Pedro
  • Barreto, Marcos
  • Denaxas, Spiros
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
[EN] Record linkage is a technique widely used to gather data stored in disparate data sources that presumably pertain to the same real world entity. This integration can be done deterministically or probabilistically, depending on the existence of common key attributes among all data sources involved. The probabilistic approach is very time-consuming due to the amount of records that must be compared, specifically in big data scenarios. In this paper, we propose and evaluate a methodology that simultaneously exploits multicore and multi-GPU architectures in order to perform the probabilistic linkage of large-scale Brazilian governmental databases. We present some algorithmic optimizations that provide high accuracy and improve performance by defining the best algorithm-architecture combination for a problem given its input size. We also discuss performance results obtained with different data samples, showing that a hybrid approach outperforms other configurations, providing an average speedup of 7.9 when linking up to 20.000 million records., This work has been partially supported by CNPq, FAPESB, Bill & Melinda Gates Foundation, The Royal Society (UK), Medical Research Council (UK), NVIDIA Hardware Grant Program, Generalitat Valenciana (Grant PROMETEOII/2014/003), Spanish Government and European Commission through TEC2015-67387-C4-1-R (MINECO/FEDER), and network CAPAP-H. We have also worked in cooperation with the EU-COST Programme Action IC1305, "Network for Sustainable Ultrascale Computing (NESUS)
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Online score-informed source separation in polyphonic mixtures using instrument spectral patterns

  • Muñoz-Montoro, A.
  • Vera-Candeas, P.
  • Cortina, Raquel
  • Combarro, E. F.
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
[EN] Soundprism is a real-time algorithm to separate polyphonic music audio into source signals, given the musical score of the audio in advance. This paper presents a framework for a Soundprism implementation. A study of the sound quality of the online score-informed source separation is shown, although a real-time implementation is not carried out. The system is compound of two stages: (1) a score follower that matches a MIDI score position to each time frame of the musical performance; and (2) a source separator based on a nonnegative matrix factorization approach guided by the score. Real audio mixtures composed of an instrumental quartets were employed to obtain preliminary results of the proposed system., Ministerio de Economía y Competitividad. Grant Number: TEC2015-67387-C4-{1, 2, 3}-R
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Real-time Soundprism

  • Muñoz-Montoro, A. J.
  • Ranilla, J.
  • Vera-Candeas, P.
  • Combarro, E. F.
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
[EN] This paper presents a parallel real-time sound source separation system for decomposing an audio signal captured with a single microphone in so many audio signals as the number of instruments that are really playing. This approach is usually known as Soundprism. The application scenario of the system is for a concert hall in which users, instead of listening to the mixed audio, want to receive the audio of just an instrument, focusing on a particular performance. The challenge is even greater since we are interested in a real-time system on handheld devices, i.e., devices characterized by both low power consumption and mobility. The results presented show that it is possible to obtain real-time results in the tested scenarios using an ARM processor aided by a GPU, when this one is present., This work has been supported by the "Ministerio de Economia y Competitividad" of Spain and FEDER under projects TEC2015-67387-C4-{1,2,3}-R.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

A pipeline structure for the block QR update in digital signal processing

  • Dolz, Manuel F.
  • Alventosa, Fran J.
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • Vidal Maciá, Antonio Manuel
[EN] There exist problems in the field of digital signal processing, such as filtering of acoustic signals that require processing a large amount of data in real time. The beamforming algorithm, for instance, is a process that can be modeled by a rectangular matrix built on the input signals of an acoustic system and, thus, changes in real time. To obtain the output signals, it is required to compute its QR factorization. In this paper, we propose to organize the concurrent computational resources of a given multicore computer in a pipeline structure to perform this factorization as fast as possible. The pipeline has been implemented using both the application programming interface OpenMP and GrPPI, a library interface to design parallel applications based on parallel patterns. We tackle not only the performance challenge but also the programmability of our idea using parallel programming frameworks., This work was supported by the Spanish Ministry of Economy and Competitiveness under MINECO and FEDER projects TIN2014-53495-R and TEC2015-67387-C4-1-R.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Generalization of the K-SVD algorithm for minimization of ß-divergence

  • García Mollá, Víctor Manuel|||0000-0003-4768-7367
  • Vidal Maciá, Antonio Manuel
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • San Juan-Sebastian, Pablo
  • Virtanen, T.
[EN] In this paper, we propose, describe, and test a modification of the K-SVD algorithm. Given a set of training data, the proposed algorithm computes an overcomplete dictionary by minimizing the ß-divergence () between the data and its representation as linear combinations of atoms of the dictionary, under strict sparsity restrictions. For the special case , the proposed algorithm minimizes the Frobenius norm and, therefore, for the proposed algorithm is equivalent to the original K-SVD algorithm. We describe the modifications needed and discuss the possible shortcomings of the new algorithm. The algorithm is tested with random matrices and with an example based on speech separation., This work has been partially supported by the EU together with the Spanish Government through TEC2015-67387-C4-1-R (MINECO/FEDER) and by Programa de FPU del Ministerio de Educacion, Cultura y Deporte FPU13/03828 (Spain).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Updating/downdating the NonNegative Matrix Factorization

  • San Juan Sebastián, Pablo
  • Vidal Maciá, Antonio Manuel
  • García Mollá, Víctor Manuel|||0000-0003-4768-7367
The Non-Negative Matrix Factorization (NNMF) is a recent numerical tool that, given a nonnegative data matrix, tries to obtain its factorization as the approximate product of two nonnegative matrices. Nowadays, this factorization is being used in many science fields; in some of these fields, real-time computation of the NNMF is required. In some scenarios, all data is not initially available and when new data (as new rows or columns) becomes available the NNMF must be recomputed. Recomputing the whole factorization every time is very costly and not suitable for real time applications. In this paper we propose several algorithms to update the NNMF factorization taking advantage of the previously computed factorizations, with similar error and lower computational cost. © 2016 Elsevier B.V. All rights reserved., This work has been partially supported by EU together with Spanish Government through TEC2015-67387-C4-1-R (MINECO/FEDER), by Generalitat Valenciana through PROMETEOII/2014/003 and by Programa de FPU del Ministerio de Educacion, Cultura y Deporte FPU13/03828 (Spain). We want to thank Dr. Pedro Vera and his team (University of Jaen) for providing us with their music analysis software.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Fast block QR update in digital signal processing

  • Alventosa, Fran J.
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • Vidal Maciá, Antonio Manuel
  • Piñero, Gema|||0000-0002-8719-8106
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] The processing of digital sound signals often requires the computation of the QR factorization of a rectangular system matrix. However, sometimes, only a given (and probably small) part of the system matrix varies from the current sample to the next one. We exploit this fact to reuse some computations carried out to process the former sample in order to save execution time in the processing of the current sample. These savings can be critical for real-time applications running on low power consumption devices with high mobility. In addition, we propose a simple out-of-order task-parallel algorithm for the QR factorization using OpenMP that exploits the multicore capability of modern processors. Furthermore, in the presence of a Graphics Processing Unit (GPU) in the system, our algorithm is able to off-load some tasks to the GPU to accelerate the computation on these hardware devices., This work was supported by the Spanish Ministry of Economy and Competitiveness under MINECO and FEDER projects TEC2015-67387-C4-1-R and TIN2014-53495-R; and the Generalitat Valenciana PROMETEOII/2014/003
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Analysis of an efficient parallel implementation of active-set Newton algorithm

  • San Juan-Sebastian, Pablo
  • Virtanen, T.
  • García Mollá, Víctor Manuel|||0000-0003-4768-7367
  • Vidal Maciá, Antonio Manuel
[EN] This paper presents an analysis of an efficient parallel implementation of the active-set Newton algorithm (ASNA), which is used to estimate the nonnegative weights of linear combinations of the atoms in a large-scale dictionary to approximate an observation vector by minimizing the Kullback¿Leibler divergence between the observation vector and the approximation. The performance of ASNA has been proved in previous works against other state-of-the-art methods. The implementations analysed in this paper have been developed in C, using parallel programming techniques to obtain a better performance in multicore architectures than the original MATLAB implementation. Also a hardware analysis is performed to check the influence of CPU frequency and number of CPU cores in the different implementations proposed. The new implementations allow ASNA algorithm to tackle real-time problems due to the execution time reduction obtained., This work has been partially supported by Programa de FPU del MECD, by MINECO and FEDER from Spain, under the projects TEC2015-67387- C4-1-R, and by project PROMETEO FASE II 2014/003 of Generalitat Valenciana. The authors want to thank Dr. Konstantinos Drossos for some very useful mind changing discussions. This work has been conducted in Laboratory of Signal Processing, Tampere University of Technology.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Distributed Affine Projection Algorithm Over Acoustically Coupled Sensor Networks

  • Ferrer Contreras, Miguel|||0000-0002-8743-1887
  • Gonzalez, Alberto|||0000-0002-6984-3212
  • de Diego Antón, María|||0000-0001-9948-3396
  • Piñero, Gema|||0000-0002-8719-8106
[EN] In this paper, we present a distributed affine projection (AP) algorithm for an acoustic sensor network where the nodes are acoustically coupled. Every acoustic node is composed of a microphone, a processor, and an actuator to control the sound field. This type of networks can use distributed adaptive algorithms to deal with the active noise control (ANC) problem in a cooperative manner, providing more flexible and scalable ANC systems. In this regard, we introduce here a distributed version of the multichannel filtered-x AP algorithm over an acoustic sensor network that it is called distributed filtered-x AP (DFxAP) algorithm. The analysis of the mean and the mean-square deviation performance of the algorithm at each node is given for a network with a ring topology and without constraints in the communication layer. The theoretical results are validated through several simulations. Moreover, simulations show that the proposed DFxAP outperforms the previously reported distributed multiple error filtered-x least mean square algorithm., This work was supported in part by EU together with Spanish Government under Grant TEC2015-67387-C4-1-R (MINECO/FEDER), and in part by Generalitat Valenciana under PROMETEOII/2014/003.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Computing Matrix Trigonometric Functions with GPUs through Matlab

  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • Peinado Pinilla, Jesús|||0000-0002-9048-5106
  • Ibáñez González, Jacinto Javier|||0000-0002-6912-4453
  • Sastre, Jorge|||0000-0002-8612-6717
  • Defez Candel, Emilio|||0000-0002-3303-6371
[EN] This paper presents an implementation of one of the most up-to-day algorithms proposed to compute the matrix trigonometric functions sine and cosine. The method used is based on Taylor series approximations which intensively uses matrix multiplications. To accelerate matrix products, our application can use from one to four NVIDIA GPUs by using the NVIDIA cublas and cublasXt libraries. The application, implemented in C++, can be used from the Matlab command line thanks to the mex files provided. We experimentally assess our implementation in modern and very high-performance NVIDIA GPUs., This work has been supported by Spanish Ministerio de Economia y Competitividad and the European Regional Development Fund (ERDF) Grants TIN2014-59294-P and TEC2015-67387-C4-1-R
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Tesis doctoral (DoctoralThesis)

Smart Sound Control in Acoustic Sensor Networks: a Perceptual Perspective

  • Estreder-Campos, Juan|||0000-0001-9728-4422
[ES] Los sistemas de audio han experimentado un gran desarrollo en los últimos años gracias al aumento de dispositivos con procesadores de alto rendimiento capaces de realizar un procesamiento cada vez más eficiente. Además, las comunicaciones inalámbricas permiten a los dispositivos de una red estar ubicados en diferentes lugares sin limitaciones físicas. La combinación de estas tecnologías ha dado lugar a la aparición de las redes de sensores acústicos (ASN). Una ASN está compuesta por nodos equipados con transductores de audio, como micrófonos o altavoces. En el caso de la monitorización acústica del campo, sólo es necesario incorporar sensores acústicos a los nodos ASN. Sin embargo, en el caso de las aplicaciones de control, los nodos deben interactuar con el campo acústico a través de altavoces. La ASN puede implementarse mediante dispositivos de bajo coste, como Raspberry Pi o dispositivos móviles, capaces de gestionar varios micrófonos y altavoces y de ofrecer una buena capacidad de cálculo. Además, estos dispositivos pueden comunicarse mediante conexiones inalámbricas, como Wi-Fi o Bluetooth. Por lo tanto, en esta tesis, se propone una ASN compuesta por dispositivos móviles conectados a altavoces inalámbricos mediante un enlace Bluetooth. Además, el problema de la sincronización entre los dispositivos de una ASN es uno de los principales retos a abordar, ya que el rendimiento del procesamiento de audio es muy sensible a la falta de sincronismo. Por lo tanto, también se lleva a cabo un análisis del problema de sincronización entre dispositivos conectados a altavoces inalámbricos en una ASN. En este sentido, una de las principales aportaciones es el análisis de la latencia de audio cuando los nodos acústicos de la ASN están formados por dispositivos móviles que se comunican altavoces mediante enlaces Bluetooth. Una segunda contribución significativa de esta tesis es la implementación de un método para sincronizar los diferentes dispositivos de una ASN, junto con un estudio de sus limitaciones. Por último, se ha introducido el método propuesto para implementar aplicaciones de zonas sonoras personales (PSZ). Por lo tanto, la implementación y el análisis del rendimiento de diferentes aplicaciones de audio sobre una ASN compuesta por dispositivos móviles y altavoces inalámbricos es también una contribución significativa en el área de las ASN. Cuando el entorno acústico afecta negativamente a la percepción de la señal de audio emitida por los altavoces de la ASN, se uti­lizan técnicas de ecualización para mejorar la percepción de la señal de audio. Para ello, en esta tesis se implementa un sistema de ecualización inteligente. Para ello, se emplean algoritmos psicoacústicos para implementar un procesamiento inteligente basado en el sis­tema auditivo humano capaz de adaptarse a los cambios del entorno. Por ello, otra contribución importante de esta tesis es el análisis del enmas­caramiento espectral entre dos sonidos complejos. Este análisis permitirá calcular el umbral de enmascaramiento de un sonido con más precisión que los métodos utilizados actualmente. Este método se utiliza para implementar una aplicación de ecualización perceptiva que pretende mejorar la percepción de la señal de audio en presencia de un ruido ambien­tal. Para ello, esta tesis propone dos algoritmos de ecualización diferentes: 1) la pre-ecualización de la señal de audio para que se perciba por encima del umbral de enmascaramiento del ruido ambiental y 2) diseñar un con­trol de ruido ambiental perceptivo en los sistemas de ecualización activa de ruido (ANE), para que el nivel de ruido ambiental percibido esté por debajo del umbral de enmascaramiento de la señal de audio. Por lo tanto, la ultima aportación de esta tesis es la implementación de una aplicación de ecualización perceptiva con los dos diferentes algorit­mos de ecualización embebidos y el análisis de su rendimiento a través del banco de pruebas realizado en el laboratorio GTAC-iTEAM., [CA] El sistemes de so han experimentat un gran desenvolupament en els últims anys gràcies a l'augment de dispositius amb processadors d'alt rendiment capaços de realitzar un processament d'àudio cada vegada més eficient. D'altra banda, l'expansió de les comunicacions inalàmbriques ha permès implementar xarxes en les quals els dispositius poden estar situats a difer­ents llocs sense limitacions físiques. La combinació d'aquestes tecnologies ha donat lloc a l'aparició de les xarxes de sensors acústics (ASN). Una ASN està composta per nodes equipats amb transductors d'àudio, com micr`ofons o altaveus. En el cas del monitoratge del camp acústic, només cal incorporar sensors acústics als nodes de l'ASN. No obstant això, en el cas de les aplicacions de control, els nodes han d'interactuar amb el camp acústic a través d'altaveus. Una ASN pot implementar-se mitjant¿cant dispositius de baix cost, com ara Raspberry Pi o dispositius mòbils, capaços de gestionar di­versos micròfons i altaveus i d'oferir una bona capacitat computacional. A més, aquests dispositius poden comunicar-se a través de connexions inalàmbriques, com Wi-Fi o Bluetooth. Per això, en aquesta tesi es proposa una ASN composta per dispositius mòbils connectats a altaveus inalàmbrics a través d'un enllaç Bluetooth. El problema de la sincronització entre els dispositius d'una ASN és un dels principals reptes a abordar ja que el rendiment del processament d'àudio és molt sensible a la falta de sincronisme. Per tant, també es duu a terme una anàlisi profunda del problema de la sincronització entre els dispositius comercials connectats als altaveus inalàmbrics en una ASN. En aquest sentit, una de les principals contribucions és l'anàlisi de la latència d'àudio quan els nodes acústics en l'ASN estan compostos per dispositius mòbils que es comuniquen amb els altaveus corresponents mitjançant enllaços Bluetooth. Una segona contribuciò sig­nificativa d'aquesta tesi és la implementació d'un mètode per sincronitzar els diferents dispositius d'una ASN, juntament amb un estudi de les seves limitacions. Finalment, s'ha introduït el mètode proposat per implemen­tar aplicacions de zones de so personal. Per tant, la implementació i l'anàlisi del rendiment de diferents aplicacions d'àudio sobre una ASN composta per dispositius mòbils i al­taveus inalàmbrics és també una contribució significativa a l'àrea de les ASN. Quan l'entorn acústic afecta negativament a la percepció del senyal d'àudio emesa pels altaveus de l'ASN, es fan servir tècniques d'equalització per a millorar la percepció del senyal d'àudio. En consequència, en aquesta tesi s'implementa un sistema d'equalització intel·ligent. Per això, s'utilitzen algoritmes psicoacústics per implementar un processament intel·ligent basat en el sistema audi­tiu humà capaç d'adaptar-se als canvis de l'entorn. Per aquest motiu, una altra contribució important d'aquesta tesi és l'anàlisi de l'emmascarament espectral entre dos sons complexos. Aquesta anàlisi permetrà calcular el llindar d'emmascarament d'un so sobre amb més precisió que els mètodes utilitzats actualment. Aquest mètode s'utilitza per a imple­mentar una aplicació d'equalització perceptual que pretén millorar la per­cepció del senyal d'àudio en presència d'un soroll ambiental. Per això, aquesta tesi proposa dos algoritmes d'equalització diferents: 1) la pree­qualització del senyal d'àudio perquè es percebi per damunt del llindar d'emmascarament del soroll ambiental i 2) dissenyar un control de soroll ambiental perceptiu en els sistemes d'equalització activa de soroll (ANE) de manera que el nivell de soroll ambiental percebut estiga per davall del llindar d'emmascarament del senyal d'àudio. Per tant, l'última aportació d'aquesta tesi és la implementació d'una aplicació d'equalització perceptiva amb els dos algoritmes d'equalització embeguts i l'anàlisi del seu rendiment a través del banc de proves realitzat al laboratori GTAC-iTEAM., [EN] Audio systems have been extensively developed in recent years thanks to the increase of devices with high-performance processors able to per­form more efficient processing. In addition, wireless communications allow devices in a network to be located in different places without physical limitations. The combination of these technologies has led to the emergence of Acoustic Sensor Networks (ASN). An ASN is com­posed of nodes equipped with audio transducers, such as microphones or speakers. In the case of acoustic field monitoring, only acoustic sensors need to be incorporated into the ASN nodes. However, in the case of control applications, the nodes must interact with the acoustic field through loudspeakers. ASN can be implemented through low-cost devices, such as Rasp­berry Pi or mobile devices, capable of managing multiple mi­crophones and loudspeakers and offering good computational capacity. In addition, these devices can communicate through wireless connections, such as Wi-Fi or Bluetooth. Therefore, in this dissertation, an ASN composed of mobile devices connected to wireless speak­ers through a Bluetooth link is proposed. Additionally, the problem of syn­chronization between the devices in an ASN is one of the main challenges to be addressed since the audio processing performance is very sensitive to the lack of synchronism. Therefore, an analysis of the synchroniza­tion problem between devices connected to wireless speakers in an ASN is also carried out. In this regard, one of the main contributions is the analysis of the audio latency of mobile devices when the acoustic nodes in the ASN are comprised of mobile devices communicating with the corresponding loudspeakers through Bluetooth links. A second significant contribution of this dissertation is the implementation of a method to synchronize the different devices of an ASN, together with a study of its limitations. Finally, the proposed method has been introduced in order to implement personal sound zones (PSZ) applications. Therefore, the imple­mentation and analysis of the performance of different audio applications over an ASN composed of mobile devices and wireless speakers is also a significant contribution in the area of ASN. In cases where the acoustic environment negatively affects the percep­tion of the audio signal emitted by the ASN loudspeakers, equalization techniques are used with the objective of enhancing the perception thresh­old of the audio signal. For this purpose, a smart equalization system is implemented in this dissertation. In this regard, psychoacous­tic algorithms are employed to implement a smart processing based on the human hearing system capable of adapting to changes in the envi­ronment. Therefore, another important contribution of this thesis focuses on the analysis of the spectral masking between two complex sounds. This analysis will allow to calculate the masking threshold of one sound over the other in a more accurate way than the currently used methods. This method is used to implement a perceptual equalization application that aims to improve the perception threshold of the audio signal in presence of ambient noise. To this end, this thesis proposes two different equalization algorithms: 1) pre-equalizing the audio signal so that it is perceived above the ambient noise masking threshold and 2) designing a perceptual control of ambient noise in active noise equalization (ANE) systems, so that the perceived ambient noise level is below the masking threshold of the audio signal. Therefore, the last contribution of this dissertation is the imple­mentation of a perceptual equalization application with the two different embedded equalization algorithms and the analysis of their performance through the testbed carried out in the GTAC-iTEAM laboratory., This work has received financial support of the following projects: • SSPRESING: Smart Sound Processing for the Digital Living (Reference: TEC2015-67387-C4-1-R. Entity: Ministerio de Economia y Empresa. Spain). • FPI: Ayudas para contratos predoctorales para la formación de doctores (Reference: BES-2016-077899. Entity: Agencia Estatal de Investigación. Spain). DANCE: Dynamic Acoustic Networks for Changing Environments (Reference: RTI2018-098085-B-C41-AR. Entity: Agencia Estatal de Investigación. Spain). • DNOISE: Distributed Network of Active Noise Equalizers for Multi-User Sound Control (Reference: H2020-FETOPEN-4-2016-2017. Entity: I+D Colaborativa competitiva. Comisión de las comunidades europea).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Accelerating Multi-Channel Filtering of Audio Signal on ARM Processors

  • Belloch Rodríguez, José Antonio
  • Alventosa, Fran J.
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
  • Vidal Maciá, Antonio Manuel
Tablets and smart phones are nowadays equipped with low-power processor architectures such as the ARMv7 and the ARMv8 series. These processors integrate powerful SIMD units to exploit the intrinsic data-parallelism of most media and signal processing applications. In audio signal processing, there exist multiple problems that require filtering operations such as equalizations or signal synthesizers, among others. Most of these applications can be efficiently executed today on mobile devices by leveraging the processor SIMD unit. In this paper, we target the implementation of multi-channel filtering of audio signals on ARM architectures. To this end, we consider two common audio filter structures: FIR and IIR. The latter is analyzed in two different forms: direct form I and parallel form. Our results show that the SIMD-accelerated implementation increases the processing speed by a factor of 4 × with respect to the original code, and our hand-tuned SIMD implementation outperforms the auto- vectorized code by a factor of 2× . These results allow us to deal in real time with multi-channel systems composed of 260 FIR filters with 256 coefficients, or 125 IIR filters with 256 coefficients, of INT16 data type., The researchers from Universitat Jaume I are supported by the CICYT projects TIN2014-53495-R and TIN2011-23283 of the Ministerio de Economia y Competitividad and FEDER. The authors from the Universitat Politecnica de Valencia are supported by projects TEC2015-67387-C4-1-R and PROMETEOII/2014/003. This work was also supported from the European Union FEDER (CAPAP-H5 network TIN2014-53522-REDT).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Optimized Fundamental Signal Processing Operations for Energy Minimization on Heterogeneous Mobile Devices

  • Belloch Rodríguez, José Antonio
  • Badia Contelles, J. M.
  • Igual Peña, Francisco Daniel
  • Gonzalez, Alberto|||0000-0002-6984-3212
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
[EN] Numerous signal processing applications are emerging on both mobile and high-performance computing systems. These applications are subject to responsiveness constraints for user interactivity and, at the same time, must be optimized for energy efficiency. The increasingly heterogeneous power-versus-performance profile of modern hardware introduces new opportunities for energy savings as well as challenges. In this line, recent systems-on-chip (SoC) composed of low-power multicore processors, combined with a small graphics accelerator (or GPU), yield a notable increment of the computational capacity while partially retaining the appealing low power consumption of embedded systems. This paper analyzes the potential of these new hardware systems to accelerate applications that involve a large number of floating-point arithmetic operations mainly in the form of convolutions. To assess the performance, a headphone-based spatial audio application for mobile devices based on a Samsung Exynos 5422 SoC has been developed. We discuss different implementations and analyze the tradeoffs between performance and energy efficiency for different scenarios and configurations. Our experimental results reveal that we can extend the battery lifetime of a device featuring such an architecture by a 238% by properly configuring and leveraging the computational resources., This work was supported by the Spanish Ministerio de Economia y Competitividad projects under Grant TIN2014-53495-R and Grant TEC2015-67387-C4-1-R, in part by the University Project UJI-B2016-20, in part by the Project PROMETEOII/2014/003. The work of J. A. Belloch was supported by the GVA Post-Doctoral Contract under Grant APOSTD/2016/069. This paper was recommended by Associate Editor Y. Ha.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Adaptive Filtered-x Algorithms for Room Equalization Based on Block-Based Combination Schemes

  • Fuster Criado, Laura
  • Azpicueta-Ruiz, Luis A.
  • de Diego Antón, María|||0000-0001-9948-3396
  • Ferrer Contreras, Miguel|||0000-0002-8743-1887
[EN] Room equalization has become essential for sound reproduction systems to provide the listener with the desired acoustical sensation. Recently, adaptive filters have been proposed as an effective tool in the core of these systems. In this context, this paper introduces different novel schemes based on the combination of adaptive filters idea: a versatile and flexible approach that permits obtaining adaptive schemes combining the capabilities of several independent adaptive filters. In this way, we have investigated the advantages of a scheme called combination of block-based adaptive filters which allows a blockwise combination splitting the adaptive filters into nonoverlapping blocks. This idea was previously applied to the plant identification problem, but has to be properly modified to obtain a suitable behavior in the equalization application. Moreover, we propose a scheme with the aim of further improving the equalization performance using the a priori knowledge of the energy distribution of the optimal inverse filter, where the block filters are chosen to fit with the coefficients energy distribution. Furthermore, the biased block-based filter is also introduced as a particular case of the combination scheme, especially suited for low signal-to-noise ratios (SNRs) or sparse scenarios. Although the combined schemes can be employed with any kind of adaptive filter, we employ the filtered-x improved proportionate normalized least mean square algorithm as basis of the proposed algorithms, allowing to introduce a novel combination scheme based on partitioned block schemes where different blocks of the adaptive filter use different parameter settings. Several experiments are included to evaluate the proposed algorithms in terms of convergence speed and steady-state behavior for different degrees of sparseness and SNRs., The work of L. A. Azpicueta-Ruiz was supported in part by the Comtmidad de Madrid through CASI-CAM-CM under Grant S2013/ICE-2845, in part by the Spanish Ministry of Economy and Competitiveness through DAMA under Grant TIN2015-70308-REDT, and Grant TEC2014-52289-R, and in part by the European Union. The work of L. Fuster, M. Ferrer, and M. de Diego was supported in part by EU together with the Spanish Government under Grant TEC2015-67387-C4-1-R (MINECO/FEDER), and in part by the Cieneralitat Valenciana under Grant PROMETEOII/2014/003. The associate editor coordinating the review of this manuscript and approving it for publication was Prof. Simon Dodo.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Automatic Tuning to Performance Modelling of Matrix Polynomials on Multicore and Multi-GPU Systems

  • Boratto, Murilo
  • Gimenez, Domingo
  • Lastovetsky, Alexey
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
[EN] Automatic tuning methodologies have been used in the design of routines in recent years. The goal of these methodologies is to develop routines which automatically adapt to the conditions of the underlying computational system so that efficient executions are obtained independently of the end- user experience. This paper aims to explore programming routines that can automatically be adapted to the computational system conditions thanks to these automatic tuning methodologies. In particular, we have worked on the evaluation of matrix polynomials on multicore and multi-GPU systems as a target application. This application is very useful for the computation of matrix functions like the sine or cosine but, at the same time, the application is very time consuming since the basic computational kernel, which is the matrix multiplication, is carried out many times. The use of all available resources within a node in an easy and efficient way is crucial for the end user., This work has been partially supported by Generalitat Valenciana under Grant PROM-ETEOII/2014/003, and by the Spanish MINECO, as well as European Commission FEDER funds, under Grant TEC2015-67387-C4-1-R and TIN2015-66972-C5-3-R, and network CAPAP-H. Also, we have work in cooperation with the EU-COST Programme Action IC1305, "Network for Sustainable Ultrascale Computing (NESUS)".
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Control Effort Strategies for Acoustically Coupled Distributed Acoustic Nodes

  • Antoñanzas-Manuel, Christian
  • Ferrer Contreras, Miguel|||0000-0002-8743-1887
  • de Diego Antón, María|||0000-0001-9948-3396
  • Gonzalez, Alberto|||0000-0002-6984-3212
[EN] This paper considers the effect of effort constraints on the behavior of an active noise control (ANC) system over a distributed network composed of acoustic nodes. A distributed implementation can be desirable in order to provide more flexible, versatile, and scalable ANC systems. In this regard, the distributed version of the multiple error filtered-x least mean square (DMEFxLMS) algorithm that allows collaboration between nodes has shown excellent properties. However, practical constraints need to be considered since, in real scenarios, the acoustic nodes are equipped with power constrained actuators. If these constraints are not considered within the adaptive algorithm, the control signals may increase and saturate the hardware devices, causing system instability. To avoid this drawback, a control effort weighting can be considered in the cost function of the distributed algorithm at each node. Therefore, a control effort strategy over the output signals at each node is used to keep them under a given threshold and ensuring the distributed ANC system stability. Experimental results show that, assuming ideal network communications, the proposed distributed algorithm achieves the same performance as the leaky centralized ANC system. A performance evaluation of several versions of the leaky DMEFxLMS algorithm in realistic scenarios is also included., This work has been supported by European Union ERDF together with Spanish Government through TEC2015-67387-C4-1-R project and Generalitat Valenciana through PROMETEOII/2014/003 project.