Sławomir Zieliński - Academia.edu (original) (raw)

Uploads

Papers by Sławomir Zieliński

Computer Information Systems and Industrial Management, 2020

This paper presents a method for augmenting the motion capture trajectories to improve generaliza... more This paper presents a method for augmenting the motion capture trajectories to improve generalization performance of recurrent long short-term memory (LSTM) neural networks. The presented algorithm is based on the interpolation of existing time series and can be applied only to segmented or easy-to-segment data due to the possibility of blending similar motion trajectories that are not significantly time-shifted. The paper shows the results of the classification efficiency with and without augmentation for two publicly available databases: Multimodal Kinect-IMU Dataset and National Chiao Tung University Multisensor Fitness Dataset. The former contains the data representing separate human computer interaction gestures, while the latter comprises the data of unsegmented series of body exercises. As a result of using the presented algorithm, the classification accuracy increased by approximately 11% points for the first dataset and 8% points for the second one.

This repository contains supplementary material for the paper titled '<em>Automatic Spa... more This repository contains supplementary material for the paper titled '<em>Automatic Spatial Audio Scene Classification in Binaural Recordings of Music</em>.' The database consists of the five following folders: Binaural recordings used for training Binaural recordings used for testing Extracted features Classification algorithm Music credits

Computer Information Systems and Industrial Management, 2020

The article presents an algorithm for classifying the style of expression of violin playing based... more The article presents an algorithm for classifying the style of expression of violin playing based on IMU sensor, located on the violinists forearm. In the initial phase of research, the original set of measured signals was extended by transferring them to new coordinate systems. Additional motion dynamics signals, including estimated linear velocity, have been obtained using transformations typical for inertial navigation systems (INS). In the next part of the work, universal features as well as indicators typical for IMU signals were extracted. The final experiment concerned the comparative effectiveness of data classification, using features selected by mutual information and random forest algorithms. The evaluation of the performance of the proposed algorithm has been carried out using a publicly available database. The obtained level of classification accuracy exceeded 90%.

Spatial audio becomes increasingly popular in domestic and mobile multimedia applications. Evalua... more Spatial audio becomes increasingly popular in domestic and mobile multimedia applications. Evaluating quality of experience (QoE) of such applications requires the development of algorithms capable of identification and quantification of perceptual characteristics of spatial audio scenes. This paper introduces a method for the automatic categorization of surround sound recordings using a criterion based on the distribution of foreground and background audio content around a listener. The principles of the method were demonstrated using a study in which a corpus of 110 five-channel surround sound recordings was computationally classified according to the two basic spatial audio scene categories. In order to develop the proposed method a novel metric, representing spatial audio characteristics, was identified. Moreover, five machine learning algorithms, including neural networks, random forests and support vector machines, were employed and their performance compared. According to the...

A systematic review of typical biases encountered in modern audio quality listening tests is pres... more A systematic review of typical biases encountered in modern audio quality listening tests is presented. The following three types of bias are discussed in more detail: bias due to affective judgments, response mapping bias, and interface bias. In addition, a potential bias due to perceptually nonlinear graphic scales is discussed. A number of recommendations aiming to reduce the aforementioned biases are provided, including an in-depth discussion of direct and indirect anchoring techniques.

2021 Immersive and 3D Audio: from Architecture to Automotive (I3DA), 2021

This paper presents a method for augmenting the motion capture trajectories to improve generaliza... more This paper presents a method for augmenting the motion capture trajectories to improve generalization performance of recurrent long short-term memory (LSTM) neural networks. The presented algorithm is based on the interpolation of existing time series and can be applied only to segmented or easy-to-segment data due to the possibility of blending similar motion trajectories that are not significantly time-shifted. The paper shows the results of the classification efficiency with and without augmentation for two publicly available databases: Multimodal Kinect-IMU Dataset and National Chiao Tung University Multisensor Fitness Dataset. The former contains the data representing separate human computer interaction gestures, while the latter comprises the data of unsegmented series of body exercises. As a result of using the presented algorithm, the classification accuracy increased by approximately 11% points for the first dataset and 8% points for the second one.

Artificial Intelligence and Soft Computing, 2018

Binaural technology becomes increasingly popular in the multimedia systems. This paper identifies... more Binaural technology becomes increasingly popular in the multimedia systems. This paper identifies a set of features of binaural recordings suitable for the automatic classification of the four basic spatial audio scenes representing the most typical patterns of audio content distribution around a listener. Moreover, it compares the five artificial-intelligence-based methods applied to the classification of binaural recordings. The results show that both the spatial and the spectro-temporal features are essential to accurate classification of binaurally rendered acoustic scenes. The spectro-temporal features appear to have a stronger influence on the classification results than the spatial metrics. According to the obtained results, the method based on the support vector machine, exploiting the features identified in the study, yields the classification accuracy approaching 84%. I.

Computer Information Systems and Industrial Management, 2021

Nota: El presente trabajo, en su totalidad o cualquiera de sus partes, no debe ser considerado co... more Nota: El presente trabajo, en su totalidad o cualquiera de sus partes, no debe ser considerado como una publicación, incluso a pesar de estar disponible sin restricciones a través de un repositorio institucional. Esta declaración se alinea con las prácticas y recomendaciones presentadas por el Committee on Publication Ethics COPE descritas por Barbour et al. (2017) Discussion document on best practice for issues around theses publishing, disponible en http://bit.ly/COPETheses. UNPUBLISHED DOCUMENT Note: The following capstone project is available through Universidad San Francisco de Quito USFQ institutional repository. Nonetheless, this project-in whole or in part-should not be considered a publication. This statement follows the recommendations presented by the Committee on Publication Ethics COPE described by Barbour et al. (2017) Discussion document on best practice for issues around theses publishing available on http://bit.ly/COPETheses.

Journal of the Audio Engineering Society, 2016

This paper provides complementary data to the review of biases in audio quality listening tests b... more This paper provides complementary data to the review of biases in audio quality listening tests by Zieliński et al. (2008) [1]. The paper presents selected illustrations of range equalizing bias, centering bias, stimulus spacing bias, contraction bias, and bias due to nonlinear properties of assessment scale. The illustrations are given in graphical form and respective discussions of biases using empirical data obtained by various researchers over the period of the past 15 years. The presented collection of illustrations along with the discussion may help the experimenters to identify potential biases affecting their data and avoid typical pitfalls in reporting the outcomes of the listening tests.

Archives of Acoustics, 2014

Let X(t), t^O, be a real Gaussian process with mean 0, stationary increments, and σ 2 (t) = E\X(t... more Let X(t), t^O, be a real Gaussian process with mean 0, stationary increments, and σ 2 (t) = E\X(t)-X(0)\ 2. Here σ 2 (t) = J°° \e m-l| 2 r 2 (l + λ 2) dH(λ), for some bounded monotone H. We summarize the main results. If the derivative H' of the absolutely continuous component of H satisfies H r {λ)>C\λ\-a-1 for all large \λ\, for some 0 < a < 2, then i) The local time φ(x, t) of the sample function exists, is jointly continuous in (x, t), and satisfies a uniform Holder condition in t of any order smaller than 1-α/2, almost surely; ii) X{t), O^t^T, nowhere satisfies a Holder condition of order greater than a/2, almost surely. If, furthermore, the sample functions are almost surely continuous, then {x : dim [t :0<t <7\ X{t) = x~\ < 1-or/2} is nowhere dense, almost surely. If, in addition, σ 2 (t) < B\t\ β 9 0<t ^T for some 0 < β < 2, then dim {t :0^t<T, X(t) = x] <,1-β/2 for all x, almost surely. If X{t) is stationary and ergodic, and a = β in the conditions above, then dim {t : tl>0, X(t) = α;} = 1a/2 for all x 9 almost surely. The theme of the preceding three papers [3], [4], and [5] is that the smoothness of the local time of a Gaussian process implies the irregularity of the sample functions. Here we continue to demonstrate this implication in a quantitative way, and sharpen some of the earlier results. The original calculations for the proof of the continuity of the local time of a Gaussian process are in [3]. The conditions were simplified and weakened, and joint continuity was proved in [5]. While not strictly comparable to those in [5], the hypotheses here are more simply stated, and the conclusions are stronger (Theorem 4.1).

Computer Information Systems and Industrial Management, 2020

This paper presents a method for augmenting the motion capture trajectories to improve generaliza... more This paper presents a method for augmenting the motion capture trajectories to improve generalization performance of recurrent long short-term memory (LSTM) neural networks. The presented algorithm is based on the interpolation of existing time series and can be applied only to segmented or easy-to-segment data due to the possibility of blending similar motion trajectories that are not significantly time-shifted. The paper shows the results of the classification efficiency with and without augmentation for two publicly available databases: Multimodal Kinect-IMU Dataset and National Chiao Tung University Multisensor Fitness Dataset. The former contains the data representing separate human computer interaction gestures, while the latter comprises the data of unsegmented series of body exercises. As a result of using the presented algorithm, the classification accuracy increased by approximately 11% points for the first dataset and 8% points for the second one.

This repository contains supplementary material for the paper titled '<em>Automatic Spa... more This repository contains supplementary material for the paper titled '<em>Automatic Spatial Audio Scene Classification in Binaural Recordings of Music</em>.' The database consists of the five following folders: Binaural recordings used for training Binaural recordings used for testing Extracted features Classification algorithm Music credits

Computer Information Systems and Industrial Management, 2020

The article presents an algorithm for classifying the style of expression of violin playing based... more The article presents an algorithm for classifying the style of expression of violin playing based on IMU sensor, located on the violinists forearm. In the initial phase of research, the original set of measured signals was extended by transferring them to new coordinate systems. Additional motion dynamics signals, including estimated linear velocity, have been obtained using transformations typical for inertial navigation systems (INS). In the next part of the work, universal features as well as indicators typical for IMU signals were extracted. The final experiment concerned the comparative effectiveness of data classification, using features selected by mutual information and random forest algorithms. The evaluation of the performance of the proposed algorithm has been carried out using a publicly available database. The obtained level of classification accuracy exceeded 90%.

Spatial audio becomes increasingly popular in domestic and mobile multimedia applications. Evalua... more Spatial audio becomes increasingly popular in domestic and mobile multimedia applications. Evaluating quality of experience (QoE) of such applications requires the development of algorithms capable of identification and quantification of perceptual characteristics of spatial audio scenes. This paper introduces a method for the automatic categorization of surround sound recordings using a criterion based on the distribution of foreground and background audio content around a listener. The principles of the method were demonstrated using a study in which a corpus of 110 five-channel surround sound recordings was computationally classified according to the two basic spatial audio scene categories. In order to develop the proposed method a novel metric, representing spatial audio characteristics, was identified. Moreover, five machine learning algorithms, including neural networks, random forests and support vector machines, were employed and their performance compared. According to the...

A systematic review of typical biases encountered in modern audio quality listening tests is pres... more A systematic review of typical biases encountered in modern audio quality listening tests is presented. The following three types of bias are discussed in more detail: bias due to affective judgments, response mapping bias, and interface bias. In addition, a potential bias due to perceptually nonlinear graphic scales is discussed. A number of recommendations aiming to reduce the aforementioned biases are provided, including an in-depth discussion of direct and indirect anchoring techniques.

2021 Immersive and 3D Audio: from Architecture to Automotive (I3DA), 2021

This paper presents a method for augmenting the motion capture trajectories to improve generaliza... more This paper presents a method for augmenting the motion capture trajectories to improve generalization performance of recurrent long short-term memory (LSTM) neural networks. The presented algorithm is based on the interpolation of existing time series and can be applied only to segmented or easy-to-segment data due to the possibility of blending similar motion trajectories that are not significantly time-shifted. The paper shows the results of the classification efficiency with and without augmentation for two publicly available databases: Multimodal Kinect-IMU Dataset and National Chiao Tung University Multisensor Fitness Dataset. The former contains the data representing separate human computer interaction gestures, while the latter comprises the data of unsegmented series of body exercises. As a result of using the presented algorithm, the classification accuracy increased by approximately 11% points for the first dataset and 8% points for the second one.

Artificial Intelligence and Soft Computing, 2018

Binaural technology becomes increasingly popular in the multimedia systems. This paper identifies... more Binaural technology becomes increasingly popular in the multimedia systems. This paper identifies a set of features of binaural recordings suitable for the automatic classification of the four basic spatial audio scenes representing the most typical patterns of audio content distribution around a listener. Moreover, it compares the five artificial-intelligence-based methods applied to the classification of binaural recordings. The results show that both the spatial and the spectro-temporal features are essential to accurate classification of binaurally rendered acoustic scenes. The spectro-temporal features appear to have a stronger influence on the classification results than the spatial metrics. According to the obtained results, the method based on the support vector machine, exploiting the features identified in the study, yields the classification accuracy approaching 84%. I.

Computer Information Systems and Industrial Management, 2021

Nota: El presente trabajo, en su totalidad o cualquiera de sus partes, no debe ser considerado co... more Nota: El presente trabajo, en su totalidad o cualquiera de sus partes, no debe ser considerado como una publicación, incluso a pesar de estar disponible sin restricciones a través de un repositorio institucional. Esta declaración se alinea con las prácticas y recomendaciones presentadas por el Committee on Publication Ethics COPE descritas por Barbour et al. (2017) Discussion document on best practice for issues around theses publishing, disponible en http://bit.ly/COPETheses. UNPUBLISHED DOCUMENT Note: The following capstone project is available through Universidad San Francisco de Quito USFQ institutional repository. Nonetheless, this project-in whole or in part-should not be considered a publication. This statement follows the recommendations presented by the Committee on Publication Ethics COPE described by Barbour et al. (2017) Discussion document on best practice for issues around theses publishing available on http://bit.ly/COPETheses.

Journal of the Audio Engineering Society, 2016

This paper provides complementary data to the review of biases in audio quality listening tests b... more This paper provides complementary data to the review of biases in audio quality listening tests by Zieliński et al. (2008) [1]. The paper presents selected illustrations of range equalizing bias, centering bias, stimulus spacing bias, contraction bias, and bias due to nonlinear properties of assessment scale. The illustrations are given in graphical form and respective discussions of biases using empirical data obtained by various researchers over the period of the past 15 years. The presented collection of illustrations along with the discussion may help the experimenters to identify potential biases affecting their data and avoid typical pitfalls in reporting the outcomes of the listening tests.

Archives of Acoustics, 2014

Let X(t), t^O, be a real Gaussian process with mean 0, stationary increments, and σ 2 (t) = E\X(t... more Let X(t), t^O, be a real Gaussian process with mean 0, stationary increments, and σ 2 (t) = E\X(t)-X(0)\ 2. Here σ 2 (t) = J°° \e m-l| 2 r 2 (l + λ 2) dH(λ), for some bounded monotone H. We summarize the main results. If the derivative H' of the absolutely continuous component of H satisfies H r {λ)>C\λ\-a-1 for all large \λ\, for some 0 < a < 2, then i) The local time φ(x, t) of the sample function exists, is jointly continuous in (x, t), and satisfies a uniform Holder condition in t of any order smaller than 1-α/2, almost surely; ii) X{t), O^t^T, nowhere satisfies a Holder condition of order greater than a/2, almost surely. If, furthermore, the sample functions are almost surely continuous, then {x : dim [t :0<t <7\ X{t) = x~\ < 1-or/2} is nowhere dense, almost surely. If, in addition, σ 2 (t) < B\t\ β 9 0<t ^T for some 0 < β < 2, then dim {t :0^t<T, X(t) = x] <,1-β/2 for all x, almost surely. If X{t) is stationary and ergodic, and a = β in the conditions above, then dim {t : tl>0, X(t) = α;} = 1a/2 for all x 9 almost surely. The theme of the preceding three papers [3], [4], and [5] is that the smoothness of the local time of a Gaussian process implies the irregularity of the sample functions. Here we continue to demonstrate this implication in a quantitative way, and sharpen some of the earlier results. The original calculations for the proof of the continuity of the local time of a Gaussian process are in [3]. The conditions were simplified and weakened, and joint continuity was proved in [5]. While not strictly comparable to those in [5], the hypotheses here are more simply stated, and the conclusions are stronger (Theorem 4.1).