Abstract:'Percussive Sound'may be considered as the audible result of objects engaging ... more Abstract:'Percussive Sound'may be considered as the audible result of objects engaging in collision, such as striking, scraping or shaking. The range of sounds created by the gamut of percussion instruments and struck found objects is vast, and the timbral variety of percussion is of interest to acoustic researchers and musicians alike.
Compared with other musical instruments, percussion is unparalleled in its diversity, and the pal... more Compared with other musical instruments, percussion is unparalleled in its diversity, and the palette of sonic and timbral possibilities is vast. Developing a means of effectively describing the diversity of percussion is the primary goal of this research, and thus the Percussive Audio Lexicon (PAL) is proposed.
The challenge of providing semantically rich information about recorded sounds – and the sources that produce them – has inspired a number of approaches and schemes seeking to address the issue. Taxonomies, ontologies and folksonomies of semantic descriptors for musical instruments, music recordings, music samples and sound effects exist in various domains including the Semantic Web, online databases and stand-alone software applications. However, none exist that exclusively address drums and percussion in the form that is proposed with PAL.
An ontology of sorts, PAL offers an approach to describing the ‘state’ of virtually any percussion instrument producing a specific sound at a moment in time, from multiple perspectives simultaneously. As a theoretical tool it can be used to learn about percussion, compare similarities and differences between sounds, and explore correlations between sound and source. In the applied environment it can be used to tag percussion samples in sound libraries with a comprehensive set of semantic descriptors.
At the heart of the lexicon lie 35 Attributes of percussion instrument morphology, excitation, spectral/temporal features, descriptive adjectives and onomatopoeia. Attributes are populated with over 2500 descriptors, organised categorically in a hierarchical tree-branch structure, and distributed across three interrelated trees called Source, Sound and Subject. Up to 81 descriptors can be selected simultaneously across any or all of the trees. Attributes that address quantifiable features of time, frequency and timbre are measured with Semantic Differential (SD) scales. Descriptors are encoded with an alphanumeric ID, which acts as a form of shorthand and notational hierarchy for the lexicon.
The lexicon’s content has been developed using a range of methodologies including direct examination of instruments, audio analyses of samples, and a listening survey conducted with volunteer participants. Results from the survey verify that using descriptors from the Subject tree are meaningful and useful. Where used, points on spectral and temporal SD scales are quantified using an approach that maps the features of analysed instruments over an approximate exponential curve.
A selection of percussion samples have successfully been tagged with descriptors and are presented in the appendices and support materials, which are also located online at http://tinyurl.com/pal-thesis-files.
‘Percussive Sound’ may be considered as the audible result of objects engaging in collision, such... more ‘Percussive Sound’ may be considered as the audible result of objects engaging in collision, such as striking, scraping or shaking. The range of sounds created by the gamut of percussion instruments and struck found objects is vast, and the timbral variety of percussion is of interest to acoustic researchers and musicians alike. However such a varied body of sounds poses challenges when attempting to effectively describe them between peers, compare sounds between each other, and catalogue samples in sound libraries. The body of research published in the acoustics literature uses terms and methods that are complex, scientific and mathematical. Conversely, simple or arbitrary ‘sounds-like’ descriptions used in everyday language are often subjective, experientially based and prone to misinterpretation. The authors seek to find a compromise between the two with a systematic, analytical approach to concisely and objectively describing all percussive sounds - and the objects that create them – that may be applied to sample libraries and databases. The approach is based on a lexicon of semantic descriptors and broadly quantized values (where appropriate) that describe percussive sounds in ways that are useful and meaningful to musicians, composers, sound designers, sound library curators and physics students. The content of the lexicon is based on a set of 39 Attributes of percussive sound, each of which contains between 2 and 100+ descriptors; in total there are approximately 600. The Attributes and their associated terminology are mostly drawn from elementary physics and musicology, with some everyday terms drawn from (English) natural language where appropriate. The lexicon is organized hierarchically into categories and sub-categories to assist end users to (a) efficiently navigate the hundreds of descriptors listed, and (b) target only those Attributes and descriptors of interest at any given time.
Abstract:'Percussive Sound'may be considered as the audible result of objects engaging ... more Abstract:'Percussive Sound'may be considered as the audible result of objects engaging in collision, such as striking, scraping or shaking. The range of sounds created by the gamut of percussion instruments and struck found objects is vast, and the timbral variety of percussion is of interest to acoustic researchers and musicians alike.
Compared with other musical instruments, percussion is unparalleled in its diversity, and the pal... more Compared with other musical instruments, percussion is unparalleled in its diversity, and the palette of sonic and timbral possibilities is vast. Developing a means of effectively describing the diversity of percussion is the primary goal of this research, and thus the Percussive Audio Lexicon (PAL) is proposed.
The challenge of providing semantically rich information about recorded sounds – and the sources that produce them – has inspired a number of approaches and schemes seeking to address the issue. Taxonomies, ontologies and folksonomies of semantic descriptors for musical instruments, music recordings, music samples and sound effects exist in various domains including the Semantic Web, online databases and stand-alone software applications. However, none exist that exclusively address drums and percussion in the form that is proposed with PAL.
An ontology of sorts, PAL offers an approach to describing the ‘state’ of virtually any percussion instrument producing a specific sound at a moment in time, from multiple perspectives simultaneously. As a theoretical tool it can be used to learn about percussion, compare similarities and differences between sounds, and explore correlations between sound and source. In the applied environment it can be used to tag percussion samples in sound libraries with a comprehensive set of semantic descriptors.
At the heart of the lexicon lie 35 Attributes of percussion instrument morphology, excitation, spectral/temporal features, descriptive adjectives and onomatopoeia. Attributes are populated with over 2500 descriptors, organised categorically in a hierarchical tree-branch structure, and distributed across three interrelated trees called Source, Sound and Subject. Up to 81 descriptors can be selected simultaneously across any or all of the trees. Attributes that address quantifiable features of time, frequency and timbre are measured with Semantic Differential (SD) scales. Descriptors are encoded with an alphanumeric ID, which acts as a form of shorthand and notational hierarchy for the lexicon.
The lexicon’s content has been developed using a range of methodologies including direct examination of instruments, audio analyses of samples, and a listening survey conducted with volunteer participants. Results from the survey verify that using descriptors from the Subject tree are meaningful and useful. Where used, points on spectral and temporal SD scales are quantified using an approach that maps the features of analysed instruments over an approximate exponential curve.
A selection of percussion samples have successfully been tagged with descriptors and are presented in the appendices and support materials, which are also located online at http://tinyurl.com/pal-thesis-files.
‘Percussive Sound’ may be considered as the audible result of objects engaging in collision, such... more ‘Percussive Sound’ may be considered as the audible result of objects engaging in collision, such as striking, scraping or shaking. The range of sounds created by the gamut of percussion instruments and struck found objects is vast, and the timbral variety of percussion is of interest to acoustic researchers and musicians alike. However such a varied body of sounds poses challenges when attempting to effectively describe them between peers, compare sounds between each other, and catalogue samples in sound libraries. The body of research published in the acoustics literature uses terms and methods that are complex, scientific and mathematical. Conversely, simple or arbitrary ‘sounds-like’ descriptions used in everyday language are often subjective, experientially based and prone to misinterpretation. The authors seek to find a compromise between the two with a systematic, analytical approach to concisely and objectively describing all percussive sounds - and the objects that create them – that may be applied to sample libraries and databases. The approach is based on a lexicon of semantic descriptors and broadly quantized values (where appropriate) that describe percussive sounds in ways that are useful and meaningful to musicians, composers, sound designers, sound library curators and physics students. The content of the lexicon is based on a set of 39 Attributes of percussive sound, each of which contains between 2 and 100+ descriptors; in total there are approximately 600. The Attributes and their associated terminology are mostly drawn from elementary physics and musicology, with some everyday terms drawn from (English) natural language where appropriate. The lexicon is organized hierarchically into categories and sub-categories to assist end users to (a) efficiently navigate the hundreds of descriptors listed, and (b) target only those Attributes and descriptors of interest at any given time.
Uploads
Papers by Robert Bell
The challenge of providing semantically rich information about recorded sounds – and the sources that produce them – has inspired a number of approaches and schemes seeking to address the issue. Taxonomies, ontologies and folksonomies of semantic descriptors for musical instruments, music recordings, music samples and sound effects exist in various domains including the Semantic Web, online databases and stand-alone software applications. However, none exist that exclusively address drums and percussion in the form that is proposed with PAL.
An ontology of sorts, PAL offers an approach to describing the ‘state’ of virtually any percussion instrument producing a specific sound at a moment in time, from multiple perspectives simultaneously. As a theoretical tool it can be used to learn about percussion, compare similarities and differences between sounds, and explore correlations between sound and source. In the applied environment it can be used to tag percussion samples in sound libraries with a comprehensive set of semantic descriptors.
At the heart of the lexicon lie 35 Attributes of percussion instrument morphology, excitation, spectral/temporal features, descriptive adjectives and onomatopoeia. Attributes are populated with over 2500 descriptors, organised categorically in a hierarchical tree-branch structure, and distributed across three interrelated trees called Source, Sound and Subject. Up to 81 descriptors can be selected simultaneously across any or all of the trees. Attributes that address quantifiable features of time, frequency and timbre are measured with Semantic Differential (SD) scales. Descriptors are encoded with an alphanumeric ID, which acts as a form of shorthand and notational hierarchy for the lexicon.
The lexicon’s content has been developed using a range of methodologies including direct examination of instruments, audio analyses of samples, and a listening survey conducted with volunteer participants. Results from the survey verify that using descriptors from the Subject tree are meaningful and useful. Where used, points on spectral and temporal SD scales are quantified using an approach that maps the features of analysed instruments over an approximate exponential curve.
A selection of percussion samples have successfully been tagged with descriptors and are presented in the appendices and support materials, which are also located online at http://tinyurl.com/pal-thesis-files.
However such a varied body of sounds poses challenges when attempting to effectively describe them between peers, compare sounds between each other, and catalogue samples in sound libraries. The body of research published in the acoustics literature uses terms and methods that are complex, scientific and mathematical. Conversely, simple or arbitrary ‘sounds-like’ descriptions used in everyday language are often subjective, experientially based and prone to misinterpretation. The authors seek to find a compromise between the two with a systematic, analytical approach to concisely and objectively describing all percussive sounds - and the objects that create them – that may be applied to sample libraries and databases.
The approach is based on a lexicon of semantic descriptors and broadly quantized values (where appropriate) that describe percussive sounds in ways that are useful and meaningful to musicians, composers, sound designers, sound library curators and physics students.
The content of the lexicon is based on a set of 39 Attributes of percussive sound, each of which contains between 2 and 100+ descriptors; in total there are approximately 600. The Attributes and their associated terminology are mostly drawn from elementary physics and musicology, with some everyday terms drawn from (English) natural language where appropriate. The lexicon is organized hierarchically into categories and sub-categories to assist end users to (a) efficiently navigate the hundreds of descriptors listed, and (b) target only those Attributes and descriptors of interest at any given time.
The challenge of providing semantically rich information about recorded sounds – and the sources that produce them – has inspired a number of approaches and schemes seeking to address the issue. Taxonomies, ontologies and folksonomies of semantic descriptors for musical instruments, music recordings, music samples and sound effects exist in various domains including the Semantic Web, online databases and stand-alone software applications. However, none exist that exclusively address drums and percussion in the form that is proposed with PAL.
An ontology of sorts, PAL offers an approach to describing the ‘state’ of virtually any percussion instrument producing a specific sound at a moment in time, from multiple perspectives simultaneously. As a theoretical tool it can be used to learn about percussion, compare similarities and differences between sounds, and explore correlations between sound and source. In the applied environment it can be used to tag percussion samples in sound libraries with a comprehensive set of semantic descriptors.
At the heart of the lexicon lie 35 Attributes of percussion instrument morphology, excitation, spectral/temporal features, descriptive adjectives and onomatopoeia. Attributes are populated with over 2500 descriptors, organised categorically in a hierarchical tree-branch structure, and distributed across three interrelated trees called Source, Sound and Subject. Up to 81 descriptors can be selected simultaneously across any or all of the trees. Attributes that address quantifiable features of time, frequency and timbre are measured with Semantic Differential (SD) scales. Descriptors are encoded with an alphanumeric ID, which acts as a form of shorthand and notational hierarchy for the lexicon.
The lexicon’s content has been developed using a range of methodologies including direct examination of instruments, audio analyses of samples, and a listening survey conducted with volunteer participants. Results from the survey verify that using descriptors from the Subject tree are meaningful and useful. Where used, points on spectral and temporal SD scales are quantified using an approach that maps the features of analysed instruments over an approximate exponential curve.
A selection of percussion samples have successfully been tagged with descriptors and are presented in the appendices and support materials, which are also located online at http://tinyurl.com/pal-thesis-files.
However such a varied body of sounds poses challenges when attempting to effectively describe them between peers, compare sounds between each other, and catalogue samples in sound libraries. The body of research published in the acoustics literature uses terms and methods that are complex, scientific and mathematical. Conversely, simple or arbitrary ‘sounds-like’ descriptions used in everyday language are often subjective, experientially based and prone to misinterpretation. The authors seek to find a compromise between the two with a systematic, analytical approach to concisely and objectively describing all percussive sounds - and the objects that create them – that may be applied to sample libraries and databases.
The approach is based on a lexicon of semantic descriptors and broadly quantized values (where appropriate) that describe percussive sounds in ways that are useful and meaningful to musicians, composers, sound designers, sound library curators and physics students.
The content of the lexicon is based on a set of 39 Attributes of percussive sound, each of which contains between 2 and 100+ descriptors; in total there are approximately 600. The Attributes and their associated terminology are mostly drawn from elementary physics and musicology, with some everyday terms drawn from (English) natural language where appropriate. The lexicon is organized hierarchically into categories and sub-categories to assist end users to (a) efficiently navigate the hundreds of descriptors listed, and (b) target only those Attributes and descriptors of interest at any given time.