More perspectives and compound techniques CS 275BMusic 254 Musical similarity Similarity studies in general Reductionist approaches Social cognition Timbral confounds Affective similarity ID: 564618
Download Presentation The PPT/PDF document "Musical Similarity:" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Musical Similarity: More perspectives and compound techniques
CS 275B/Music 254Slide2
Musical similaritySimilarity studies in general
Reductionist approaches
Social
cognitionTimbral confoundsAffective similarityCognitive distance metricsCompound search techniques
2016 Eleanor Selfridge-Field
2
4/11/2016Slide3
Cognitive distance metric (1)
1.
Basic Pitch-Accent Structure
Range = 0-4
A.
If meter matches target
Max = 1.00
and If subunit (e.g. quarter note) is the same Score = 1.00 orIf subunit is different (e.g., 4/8 vs. 2/4) Score = 0.50 ElseScore = 0.00 B. Percentage of matched pitches on primary beats* Max = 2.00 If matching number of scale degrees=100% Score = 2.00 orIf matching number of scale degrees =>90% Score = 1.33 orIf matched number of notes/unit =>80% Score=0.67Score = 0.67 ElseScore = 0.00 C. Percentage of matched pitches on secondary beats Max = 1.00 If matching number of scale degrees=100% Score = 1.00 orIf matching number of scale degrees=>90%Score = 0.67 orIf matched number of notes/unit =>80% Score=0.33Score = 0.33 ElseScore = 0.00
2016 Eleanor Selfridge-Field
3
4/11/2016Slide4
Cognitive distance metric (2)
II.
Basic Harmonic-Accent Structure
Range = 0-6A. Mode of work (major, minor, other)
Max = 1.00
If modes match
Score = 1.00
ElseScore = 0.00 B. Percentage of matched chords on downbeat**Max = 2.50 If unambiguous matches on primary beats =>90% Score = 2.50 or If unambiguous matches on primary beats =>80%Score = 2.00 orIf unambiguous matches on primary beat =>70%Score = 1.50 ElseScore = 0.00 C. Percentage of matched chords on secondary beats** Max = 2.00 If unambiguous matches =>90% Score = 2.00 orIf unambiguous matches =>80%Score = 1.50 orIf unambiguous matches =>70%Score = 1.00 ElseScore = 0.00 D. Percentage of matched chords on tertiary beatsMax = 0.50 If unambiguous matches =>90%Score = 0.50 Else Score = 0.02016 Eleanor Selfridge-Field44/11/2016Slide5
Cognitive distance metric (3)
Example
Pitch-Accent score
Harmonic-Accent score
Total score (additive)
Raw
Ranked
RawRankedRawRanked2a3.6725.539.1722b3.6725.048.6732c2.6766.018.6732d1.1794.556.6782e2.6764.096.6782f2.3384.556.8372g1.00102.0113.00112h3.5044.558.0062i4.0014.558.5052j 1.00104.095.00102k3.33 56.0 19.3312016 Eleanor Selfridge-Field54/11/2016Slide6
Evaluating search viability and efficiencyKrumhansl, 2000 [theory/experiment
]
Sapp, Liu,
Selfridge-Field, 2004 [practical]“Search effectiveness measures for symbolic music queries in very large databases” ISMIR 2004:http://ismir2004.ismir.net/proceedings/p051-page-266-paper135.pdf 2016 Eleanor Selfridge-Field6
4/11/2016Slide7
Sapp, Liu, Selfridge-Field (ISMIR 2004)
2016 Eleanor Selfridge-Field
7
Data
Search Effectiveness (1)
4/11/2016Slide8
Search Effectiveness (2)
2016 Eleanor Selfridge-Field
8
Pitch features
Meter features
4/11/2016Slide9
Search Effectivesness (3)
2016 Eleanor Selfridge-Field
9
Sample search
Coupled
search
4/11/2016Slide10
Results2016 Eleanor Selfridge-Field
10
4/11/2016Slide11
Sapp, Liu, Selfridge-Field (ISMIR 2004)
2016 Eleanor Selfridge-Field
11
Data
Search Effectiveness (1)
4/11/2016Slide12
Repertory
Repeated pitch intervals
Up intervals
Down intervals Symbols/theme
Total incipits
15-16th cent/Latin
29,916
66,026
67,15118,94715th-16th cent/Polish57,006130,030157117601617-18th cent/US RISM110,478344,621399,07955,49118th-19 cent/classical19,24180,16686,43010,722Essen Europe43,58153,03358,8156,232Essen Asia16,44154,57765,6842,240Luxembourg4,3559,72011,927612281,018738,173846,203100,260Stats from 2004 study4/11/20162016 Eleanor Selfridge-Field12Slide13
N
ormalized pitch usage by repertory
4/11/2016
2016 Eleanor Selfridge-Field
13Slide14
Form: Boundary strengthMarkus Pearce, Daniel Müllensiefen, Geraint Wiggins (Goldsmith): et al: “Melodic Grouping in Music Information Retrieval”
2016 Eleanor Selfridge-Field
14
After Frankland and Cohen (2004)4/11/2016Slide15
Binary (?) melodic segmentationMelodic segmentation (binary~)Pearce’s “ground truth” tests using
Essen ERK
data with binary segmentation(s)
Marcus T. Pearce, Daniel Müllensiefen, and Geraint A. Wiggins, “Melodic Grouping and Music Information Retrieval” (2010) [same], “A Comparison of Statistical and Rule-Based Models and Melodic Segmentation”, ISMIR (2008). 2016 Eleanor Selfridge-Field15
Pearce results (Springer
Verlag
, 2010)
GPR2a preference rules work
Temperley 2001 also usefulSynthesis of Grouper, LBDM, GPR2a, and Thom better that any one individuallyUnsupervised learned performed better than any statistical method4/11/2016Slide16
Pearce et al software and articlesIdyom software: https://code.soundsoftware.ac.uk/projects/idyom-project
Perceptual segmentation of melodies:
http://www.doc.gold.ac.uk/~
mas03dm/papers/icmpc08_PearceMullensiefenWiggins.pdfExpectation in audio boundaries:http://www.ncbi.nlm.nih.gov/pubmed/211803584/11/20162016 Eleanor Selfridge-Field16