Analisis aplausos

Solo disponible en BuenasTareas
  • Páginas : 16 (3879 palabras )
  • Descarga(s) : 0
  • Publicado : 13 de febrero de 2012
Leer documento completo
Vista previa del texto
Proc. of the 11th Int. Conference on Digital Audio Effects (DAFx-08), Espoo, Finland, September 1-4, 2008

INFERRING THE HAND CONFIGURATION FROM HAND CLAPPING SOUNDS Antti Jylhä and Cumhur Erkut Dept. of Signal Processing and Acoustics, Helsinki University of Technology, TKK Espoo, Finland,
ABSTRACT In this paper, a technique for inferring theconfiguration of a clapper’s hands from a hand clapping sound is described. The method was developed based on analysis of synthetic and recorded hand clap sounds, labeled with the corresponding hand configurations. A naïve Bayes classifier was constructed to automatically classify the data using two different feature sets. The results indicate that the approach is applicable for inferring the hand configuration.1. INTRODUCTION Humans are used to interact with each other by sound, and our everyday listening skills are well-developed for extracting information from our environment [1]. Similarly, in a fluent sonic interaction between a human and a computer, the computer must be able to recognize the sonic control signals the user invokes, and to distinguish them from other sounds in the environment. Such asonic interaction may occur also using everyday sounds as the conveyor of information instead of speech or music. In this paper, we use hand claps as a test case for the feasibility of such an interaction. Hand claps are a ubiquitously familiar phenomenon and easy to capture by relatively cheap equipment. Therefore, they could be widely applied in different applications. Automatic recognition ofthe hand clap type would be interesting in human-computer interaction not only because it would enable more ways of exploiting hand claps as conveyors of information, but also because it can potentially allow personified control interfaces and clapper identification. Assuming that the hand clap sounds of individual clappers are systematically different, they could be applied to control applications,which are only desired to be controlled by a specific person. In this paper, we will discuss some possibilities of estimating the system-specific parameters from the audio signals for a hand clap model. We focus on offline methods in this exploratory research. It is also a long-term objective in our research to make an online algorithm for estimating the parameters of a physics-based sound synthesissystem. 2. PERCEPTION, ANALYSIS, AND SYNTHESIS OF HAND CLAPS Hand claps are a relatively primitive conveyor of sonic information, yet they are widely applied for different purposes [2]. In different cultures hand claps are used in a musical context, and we are used to give feedback of a performance by applause [3], by indicating different levels of enthusiasm to the performers. Hand claps are anessential part of flamenco music, in which rhythmic patterns of soft and sharp claps are used as an accompaniment. Hand claps have also been used to call for service. Previous work on hand clap sounds and human-computer interaction includes for example a hand clap language as a common means of communication between humans and robots [4]. This implementation does not consider different hand claptypes, however. A recent work has also investigated the identification of synchronous vs. asynchronous applause using Mel-frequency cesptral coefficients and a genetic algorithm [5]. As sound events, hand claps of an individual are very short in time. In anechoic conditions, a hand clap sound lasts typically around 5 ms, and it is difficult to pinpoint any systematic differences between different kind ofhand claps. However, according to Repp [2], human observers are able to deduce the hand configuration of a clapper with good accuracy in a single-clapper setting by listening to the hand clap sound. Based on spectral analysis, Repp has proposed eight different hand configurations which have audible differences. These clapping modes are presented in Fig. 1.

Figure 1: Hand clap types reproduced...
tracking img