Modelling the Microphone-Related Timbral Brightness of Recorded Signals

Andy Pearce,Russell Mason,Tim Brookes

doi:10.3390/app11146461

Abstract

Brightness is one of the most common timbral descriptors used for searching audio databases, and is also the timbral attribute of recorded sound that is most affected by microphone choice, making a brightness prediction model desirable for automatic metadata generation. A model, sensitive to microphone-related as well as source-related brightness, was developed based on a novel combination of the spectral centroid and the ratio of the total magnitude of the signal above 500 Hz to that of the full signal. This model performed well on training data (r = 0.922). Validating it on new data showed a slight gradient error but good linear correlation across source types and overall (r = 0.955). On both training and validation data, the new model out-performed metrics previously used for brightness prediction.

Highlights

Audio database searching can be facilitated by metadata relating to the characteristics of each audio excerpt
Automatic generation of metadata relating to timbre can facilitate searching of audio databases, and brightness is one of the most commonly used timbral descriptors; it is the timbral attribute most affected by microphone choice
Models have been developed previously to predict the differing brightnesses of recorded sounds that result from differences between sound sources

Summary

Introduction

Audio database searching can be facilitated by metadata relating to the characteristics of each audio excerpt. A key characteristic of sound is its timbre, and brightness is one of the most commonly used timbral descriptors, e.g., it is in the top 3 most-searched timbral attributes on the Freesound audio sample repository [1]. Identified 31 perceptual attributes of recorded sound that can be affected by microphone choice. The work documented in this paper aimed to develop a model sensitive to microphone-related, as well as source-related, brightness differences and to evaluate it against existing brightness models found to be useful in other application areas, e.g., differentiating between musical instruments [3,4,5], between variations of a single instrument [6], between musical performances [7], or between vocal articulations [8]

Objectives

Discussion

Conclusion