AES New York 2015
Paper Session P14
P14 - Perception—Part 3
Saturday, October 31, 9:00 am — 12:00 pm (Room 1A07)
Chair:
Agnieszka Roginska, New York University - New York, NY, USA
P14-1 Spatial Sound Attributes—Development of a Common Lexicon—Nick Zacharov, DELTA SenseLab - Hørsholm, Denmark; Torben Holm Pedersen, DELTA SenseLab - Hørsholm, Denmark
Sound quality and spatial sound have been topics of research for decades in relation to loudspeakers and headphones as well as performance spaces (e.g., concert halls). Attributes may be used as a means to characterize sound quality through listening tests. Attribute development protocols are well reported and have been employed to a wide range of spatial sound applications. However, the usage of attribute often leads to researchers discussing the merits of the attributes as opposed to focusing upon the object of the research. Over the last few decades a large number of articles have included the development of a spatial sound attribute. This paper describes the collection of many known research articles on spatial sound attributes from a wide range of domains. As opposed to repeating traditional word elicitation and group discussion, we have chosen to use a semantic text data mining approach to find common attribute meanings, which is then followed by a sorting and refinement process with expert assessors. This process is defined in detail and the results of the semantic text mining are presented as part of the further development of a sound wheel for sound reproduction.
Convention Paper 9436 (Purchase now)
P14-2 Towards a MATLAB Toolbox for Imposing Speech Signal Impairments Following the P.TCA Schema—Friedemann Köster, University of Technology Berlin - Berlin, Germany; Falk Schiffner, University of Technology Berlin - Berlin, Germany; Dennis Guse, University of Technology Berlin - Berlin, Germany; Jens Ahrens, University of Technology Berlin - Berlin, Germany; Janto Skowronek, University of Technology Berlin - Berlin, Germany; Sebastian Möller, University of Technology Berlin - Berlin, Germany
In this paper we present and validate a freely available MATLAB Toolbox for imposing speech signal impairments similar to those occurring in real-world telecommunication systems. The purpose of the toolbox is to facilitate research on the perception of different dimensions of speech quality and their relation to technical system properties. In that context the International Telecommunication Union (ITU) is working on the annotation method P.TCA, which enables expert listeners to identify the technical cause for an observed speech signal impairment. Our contribution addresses one current challenge of P.TCA: it was found out that providing written definitions of speech degradations without exemplary listening material is not sufficient to be reliably understood by annotators. To address this issue and make the schema accessible for a wide range of users, this paper describes a systematic approach to generate and validate such exemplary listening material. A validation experiment shows that experts can identify more than half of the processed examples correctly and it encourages further research towards improving the P.TCA procedure as well as the processing algorithms.
Convention Paper 9437 (Purchase now)
P14-3 The Influence of Dumping Bias on Timbral Clarity Ratings—Kirsten Hermes, University of Surrey - Guildford, Surrey, UK; Tim Brookes, University of Surrey - Guildford, Surrey, UK; Christopher Hummersone, University of Surrey - Guildford, Surrey, UK
When listening test subjects are required to rate changes in a single attribute, but also hear changes in other attributes, their ratings can become skewed by “dumping bias.” To assess the influence of dumping bias on timbral “clarity” ratings, listeners were asked to rate stimuli: (i) in terms of clarity only; and (ii) in terms of clarity, warmth, fullness, and brightness. Clarity ratings of type (i) showed (up to 20%) larger interquartile ranges than those of type (ii). It is concluded that in single-attribute timbral rating experiments, statistical noise—potentially resulting from dumping bias—can be reduced by allowing listeners to rate additional attributes either simultaneously or beforehand.
Convention Paper 9438 (Purchase now)
P14-4 Method for Objective Evaluation of Nonlinear Distortion—Mikhail Pahomov, LG Electronics, Inc. - St. Petersburg, Russia; Yong Hyuk Na, Sr., LG Electronics, Inc. - Seoul, South Korea
A perceptual method is presented for assessing nonlinear distortion audibility in sound systems with high distortion levels of the original signals (mobile devices). The method is based on the Perceptual Evaluation of Audio Quality (PEAQ) standard [1]. To estimate the audibility of non-linear distortion, generating a content-dependent multitone signal and extracting a distortion signal from it is proposed. Then, the distortion signal’s properties are measured. Next, a regression analysis is applied to combine the properties to derive a metric that denotes the overall audible harmonic distortion. Experimental results on mobile handsets are provided to verify the high accuracy of the method.
Convention Paper 9439 (Purchase now)
P14-5 Subjective and Objective Measurements of Speech Loudness in Hands-Free Telephony—Toward an Extended Loudness Model for Telephonometry—Idir Edjekouane, Orange Labs - Lannion Cedex, France; LMA-CNRS; Cyril Plapous, Orange Labs - Lannion Cedex, France; Catherine Quinquis, Orange Labs - Lannion Cedex, France; Sabine Meunier, Centre National de la Recherche Scientifique - Marseille Cedex, France
The loudness rating technique is widely used in telephony. This technique shows some limitations with the recent advances in telecommunications. This paper proposes a new alternative for the loudness rating technique using an extension of Zwicker’s loudness model. We first investigated the loudness of speech transmitted via a telephone system and the ability of Zwicker’s model to predict the perceived loudness. The model predicts the main trends observed in perceptual data. However, a bias exists between the prediction and the measure that depends on sound pressure level. Based on our perceptual data and on recent studies, we proposed a modification of the model at the specific loudness calculation stage. This modification brought a significant improvement on the predictions.
Convention Paper 9440 (Purchase now)
P14-6 Investigation on the Phantom Image Elevation Effect—Hyunkook Lee, University of Huddersfield - Huddersfield, UK
Listening tests have been carried out in order to evaluate the phantom image elevation effect depending on horizontal stereophonic base angle. Seven ecologically valid sound sources as well as four noise sources were tested. Subjects judged the perceived image positions of phantom center image created with seven loudspeaker base angles. Results generally showed that perceived images were elevated from front to above as the loudspeaker base angle increased up to around 180°. This tendency depended on the spectral characteristics of sound source. The perceived results are explained from both physical and cognitive points of view.
Convention Paper 9441 (Purchase now)
