Timbre: Acoustics, Perception, and Cognition

Kai Siedenburg, Charalampos Saitis, Stephen McAdams

The goal of this chapter is to provide a roadmap and context for the whole volume. Hence, the chapter will provide a conceptual introduction to the notion of timbre and a brief survey of important steps in the history of timbre research. The content of the subsequent chapters will be briefly outlined and situated in an interdisciplinary framework, followed by a discussion of important research questions on timbre.

PART I: PRINCIPAL PERCEPTUAL PROCESSES

2. The Perceptual Representation of Timbre

Stephen McAdams

This chapter covers the current state of knowledge about the structure of timbre's perceptual representation. It discusses dimensional models of timbre based on multidimensional scaling (MDS) of timbre dissimilarity ratings and addresses various extensions of MDS models as well as psychophysical explanations in terms of acoustical correlates of perceptual dimensions. It further covers research on the covariance of timbre, pitch, and loudness and discusses the ways in which this covariance affects the recognition and identification of sound sources. It further outlines the utility of considering high-dimensional acoustic representations such as modulation spectra as an acoustic basis for timbre modeling.

3. Timbre Categorization and Recognition

Trevor Agus, Clara Suied, Daniel Pressnitzer

There have been many important and intriguing empirical findings on the categorization and recognition of sounds in the last seven years. This chapter reviews these studies and specifically examines the minimal amount of acoustic and temporal information required to recognize sounds such as repeated noise bursts, isolated instrument sounds, or polyphonic musical textures. The chapter thus addresses the core question regarding the timbre cues utilized by humans for the recognition of various classes of sounds.

4. Memory for Timbre

Kai Siedenburg, Daniel Müllensiefen

This chapter discusses research on long- and short-term memory for timbre. A guiding question is whether timbre is stored independently from other mental tokens (e.g., pitch as in musical melodies, or words as in verbal utterances), and whether it is governed by the same principles as those observed in these neighboring domains. Finding answers to these question will involve decomposing memory for timbre into cognitive processes such as perceptual similarity, chunking, semantic encoding, as well as accounting for the factor of auditory expertise.

5. Concepts and Semantics in Timbre Perception

Charalampos Saitis, Stefan Weinzierl

Bruno Giordano

This chapter reviews recent findings regarding the neural basis of timbre information processing from studies using both animal models as well as human brain imaging. This includes the specific neural correlates of spectral and temporal shape discrimination, findings regarding the cortical representation of spectrotemporal information, and more general models of sound source identity processing in cortex.

PART II: SPECIFIC SCENARIOS

7. Voice Processing and Voice Identity Recognition

Samuel Mathias, Katharina von Kriegstein

Humans effortlessly extract a wealth of information from speech sounds, including semantic, emotional, and details related to speaker identity. The chapter reviews the basic principles of human vocal production, behavioral studies on the processing and recognition of familiar and unfamiliar voices, as well as neural mechanisms and models of speaker recognition. The chapter further introduces to phonagnosia, the deficit of not being able to recognize familiar people by their voices and discusses its relation to autism spectrum disorder.

8. Timbre as a Structuring Force in Music

Stephen McAdams

Timbre plays a substantial part in shaping the perceptual experience of music. This chapter reviews the processes that may serve as the basis of this phenomenon with a particular focus on auditory scene analysis principles. Specific perceptual processes to be addressed include timbre's dependence on concurrent grouping (including timbral blend), the processing of sequential timbral relations, its role in sequential and segmental grouping, and the contribution of these grouping processes to musical structuring. The discussion draws from psychophysical studies as well as a discussion of selected musical examples from the Western orchestral repertoire.

9. Timbre Perception with Cochlear Implants

Jeremy Marozeau

The chapter outlines timbre perception of patients with severe or profound hearing loss that have received a cochlear implant (CI). Whereas the perception of speech in quiet works relatively well for CI patients, music perception or voice identity perception still pose great problems. The chapter will discuss CI research on timbre dissimilarity perception, musical instrument identification, auditory stream segregation, issues in voice identity and gender recognition, as well as potential improvements for CI coding strategies.

10. Timbre, Sound Quality, and Sound Design

Guillaume Lemaitre, Patrick Susini

This chapter focusses on the role of timbre in the evaluation of product sounds, related to the question of how sounds contribute to the aesthetic, functional, and emotional aspect of a product.

Research in this domain has utilized multidimensional scaling in conjunction with acoustic-descriptor-based approaches and regression modeling in order to develop models of sound quality that can be applicable in sound design. Example cases are diverse and include products such as car horns, wind turbines or consumer electronics devices such as printers. Implications for approaches to sonic interaction design are discussed.

PART III: ACOUSTICAL MODELING

Geoffroy Peeters

A sophisticated machinery for the acoustic description of sounds is essential for progress in timbre perception and cognition. This chapter provides a survey of approaches and current trends to acoustic characterization of timbre. Many scalar or time-varying descriptors are based on the Short-Time Fourier Transform from which summary measures are computed (e.g., the spectral center of gravity). Others are inspired by signal transformations that more closely follow auditory physiology. An important aspect of audio representation is that descriptors are usually highly correlated, so appropriate statistical approaches must be followed to cope with collinearity to effectively use these descriptors in the service of psychophysical explanations.

12. Modulation Representations for Speech and Music

This chapter outlines recent advances in the application and study of spectrotemporal modulation representations in speech and music. This work develops a neuro-computational framework based on spectrotemporal receptive fields, recorded from neurons in the mammalian primary auditory cortex, as well as from simulated cortical neurons. The chapter discusses the utility of applying this framework to the automatic classification of musical instrument sounds and to robust detection of speech in noise.

13. Analysis-Synthesis Approaches: Impact Sounds, Textures, and Musical Instruments

Sølvi Ystad, Mitsuko Aramaki, Richard Kronland-Martinet

This chapter introduces an analysis-synthesis framework that derives intuitive control parameters of electronic sound synthesis directly from the statistics of input sounds. The framework is based on the distinction of action and object properties that are related to the mode of sound source excitation and resonance properties, respectively. The chapter reviews recent applications of this framework to the synthesis of impact sounds, textures, and musical instrument sounds.