Upload
others
View
11
Download
1
Embed Size (px)
Citation preview
Computational modeling for
Hindustani music
Preeti Rao
Department of Electrical Engineering
Indian Institute of Technology Bombay
CompMusic Workshop, January, 2012,
KIIT College of Engineering, Gurgaon 1
Topics Covered
• What we can do with our computing tools…
• Why it may be useful to
– Musicological studies
– Pedagogy
– Access and enjoyment of music
2
Department of Electrical
Engineering , IIT Bombay
Sound and Sensation
Primary sensations Physical correlates
– loudness intensity
– pitch fundamental frequency
– timbre (“quality” ) spectro-temporal properties
Music conceptsBasic dimensions of music are melody, rhythm, harmony
and timbre.
• Melody, harmony -> based on pitch content
• Rhythm -> based on timing information
• Timbre -> relates to instrumentation, texture
A representation of these high-level attributes can be
obtained from pitch and timing information extracted by
audio signal analysis.
5
Melodic Pitch Contours• Differences in the two melodic styles are observed by the
visual comparison of the pitch contour segments
6
Pitch contour of alaap segments of Carnatic vocalist Nagavalli Nagaraj for raagSubhapanthuvarali. The y-axis denotes the pitch in cents with respect to the A-440Hz.
Discrete notes?
7
Melodic contour: applications
• Musicological studies such as intonation
• Melodic phrase matching for pedagogy: both for “emulation” and improvisation.
• Melodic similarity for music classification and retrieval including visualization.
• Transcription to aid studies deriving from symbolic notation of music
8
• Intonation (shruti is all important. Should be
accurate for given raga, style…….)
• Pitch class distributions (past work)
FRSM 2012 - 18-19 January, 2012, KIIT
College of Engineering, Gurgaon9
Ornamentation• Style, emotions, gharana (school/style) characteristics,
raga characteristics and even the personal characteristics influence the ornaments.
• Ornaments enhance the basic melodic contour and contribute to musicality/ expressiveness.
e.g. grace note, a glide between two notes, multiple oscillations of a single note, oscillation between notes….
• Of the ornaments, those that are often transcribed in notation are Meend, Andolan, Khatka, Murki, Gamak, Zamzama
[ITC SRA site: http://www.itcsra.org/alankar/alankar.html]
June 20th, 2011 DAP Lab., Dept. of EE 10 of 50
Visualization and pedagogy
DAP Lab., Dept. of EE
Model
Singer
Trained
Singer
Amateur
Singer
Time (sec)
Pitch
Fre
qu
en
cy(H
z)
Fre
qu
en
cy(H
z)
11 of 50June 20th, 2011
Observed differences
DAP Lab., Dept. of EE
• The “notes” rendered by both, trained and
amateur singers, are similar to the reference
• The differences lie between the notes…
Model Singer Trained Singer Amateur Singer
Gamak: Oscillations-on-Glide
• Periodic oscillations riding on a glide-like transition from one note to another (may or may not be of uniform amplitude)
• Gamak in Indian classical music as defined in [SRAsite] : “A gamak can be defined as a fast meend (spanning 2-3 notes normally) delivered with deliberate force and vigor and repeated in an oscillatory manner”
• Possible attributes:
– Glide - overall monotonic trend
– Oscillation - pure vibration around
the glide
• Amplitude
• Rate
DAP Lab., Dept. of EE 13 of 50
0 0.1 0.2 0.3 0.4-400
-300
-200
-100
0
100
200
300
Time (sec)
Pitch F
req (
cents
)
Objective measure: distance in parameter space...
0 20 40 60 80 100 1200
0.2
0.4
0.6
0.8
1
Singers
Ho
listic O
bje
ctiv
e S
co
re
Good
Medium
Bad
14 of 50
Correlation with perceived quality
[C. Gupta and P. Rao, "An objective evaluation tool for ornamentation in singing",
Proc. of International Symposium on Computer Music Modeling and Retrieval
(CMMR) and Frontiers of Research on Speech and Music (FRSM), 2011.]
F. Gouyon, S. Dixon
ISMIR 2006 Tutorial: Computational rhythm description
International Conference on Music Information Retrieval, 2006
Rhythm representation
Rhythm
16
o Rhythm refers to the periodic and hierarchic framework
that embeds the timing of events (onsets) within the audio
signal.
o Rhythm detection involves detecting events at each
metrical level.•Tatum
•Tactus
•Measure
o Onsets are detected via abrupt increases in loudness, or
equivalently, abrupt energy changes within frequency bands.
Rhythm is represented by the detected periodicities of the
sequence of onsets.