mazurka project update
DESCRIPTION
Mazurka Project Update. Craig Stuart Sapp CHARM Symposium Kings College, University of London 26 January 2006. Data Entry Update. Data Entry. Data Analysis. Source material: mazurka recordings. 29 performances:. 1,374 recordings of 49 mazurkas. = 28 performances/mazurka on average. - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/1.jpg)
Mazurka Project Update
Craig Stuart Sapp
CHARM SymposiumKings College, University of London
26 January 2006
![Page 2: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/2.jpg)
Data Entry Update
Data Entry Data Analysis
![Page 3: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/3.jpg)
Source material: mazurka recordings
29 performances:
• 1,374 recordings of 49 mazurkas
• 65 performers, 73 CDs
= 28 performances/mazurka on average
number of mazurka performancesin each decade
![Page 4: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/4.jpg)
Waveform
• Where are the notes?• How many notes are there?• Where are the beats?
• Reverse conducting (corrected) added to waveform
![Page 5: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/5.jpg)
MzPowerCurve
• Started development a few months before first SV release.• By-product of looking at how to extract note loudnesses from audio.• Some notes become easy to see.• Some notes obscured – mostly by beating between harmonics.
http://sv.mazurka.org.uk/MzPowerCurve
![Page 6: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/6.jpg)
MzSpectralFlux
• Implementation of Spectral Flux as described by Simon Dixon:Dixon, Simon. "Onset detection revisited" in the Proceedings of the 9th International Conference on Digital Audio Effects (DAFx'06). Montreal, Canada; September 18-20, 2006.
http://sv.mazurka.org.uk/MzSpectralFlux
• Component of the MATCH program.• Similar to the power curve idea, but measurements done on the spectrum.• Only frequency bins gaining energy are considered.• Gets rid of ½ of the harmonic beating problem.
![Page 7: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/7.jpg)
Spectral flux peak finding
• Too many false positives:
• A few false positives:
Sensitive to parameter settings:
![Page 8: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/8.jpg)
Spectral difference
• Blue vertical lines mark automatically identified beat locations.• Pink vertical lines are human-identified beat locations.• Notice relation between blue lines and pink lines.
• Tempo Tracker plugin from QUML C4DM uses same technique as Spectral Flux• But called “Spectral Difference”• Onset or Difference function not available as outputs from the plugin• Only beat locations, shown as blue lines:
![Page 9: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/9.jpg)
MzAttack
http://sv.mazurka.org.uk/MzAttack
• Developed July 2006 after last colloquia• Clear indications of note onsets• Noise peaks are difficult to separate from onset peaks, so usually used in conjuction with MzPowerCurve.• Allows for precise manual correction of reverse conducting to go from ~6 hours/performance to ~1 hour/performance.
![Page 10: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/10.jpg)
MzSpectralReflux
http://sv.mazurka.org.uk/MzSpectralReflux
• Update on the MzAttack technology based on studying Spectral Flux.• Very low noise due to harmonic beating, • Only noise left is from clicks, pops, etc., and non-musical sounds in audio.• Slightly less sensitive to parameter settings than spectral flux.• Working on reverse conducting correction time on the order of ~15-30 minutes/performance (compared to current ~1 hour/performance).
http://mazurka.org.uk/cgi-bin/tapsnap = Move taps to nearest onset
![Page 11: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/11.jpg)
Peek Under the Hood (Bonnet)
![Page 12: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/12.jpg)
Possible additions
example: {2, 3}
arithmetic mean = (2 + 4) / 2 = 3geometric mean = sqrt(2 * 4) = 2.8
spectral flatness = 2.8 / 3 = 0.94
Used to distinguish between noise and pitched sound
note noise note note
(MzSpectralFlatness)
![Page 13: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/13.jpg)
Performance data extraction
Reverse conducting
Align taps to beats
Automatic feature extraction
tempo by beat
off-beat timings
individualnote timings
individual note loudnesses
• Listen to recording and tap to beats.• Tap times recorded in Sonic Visualiser by tapping on computer keyboard.
• Reverse conducting is real-time response of listener, not actions of performer.• Adjust tap times to correct beat locations.• A bit fuzzy when RH/LH do not play in sync, or for tied notes.
![Page 14: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/14.jpg)
Data Entry Update
Data Entry Data Analysis
![Page 15: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/15.jpg)
Dynamics & Phrasing
1
2
3
all at once:
rubato
![Page 16: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/16.jpg)
Average tempo over time• Performances of mazurkas slowing down over time:
Friedman 1930
Rubinstein 1966 Indjic 2001
• Slowing down at about 3 BPM/decade
Laurence Picken, 1967: “Centeral Asian tunes in the Gagaku tradition” in Festschriftfür Walter Wiora. Kassel: Bärenreiter, 545-51.
![Page 17: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/17.jpg)
Average Tempo over time (2)
• The slow-down in performance tempos is unrelated to the age of the performer
![Page 18: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/18.jpg)
Tempo graphs
http://mazurka.org.uk/ana/tempograph
![Page 19: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/19.jpg)
Mazurka Meter
• Stereotypical mazurka rhythm:
• First beat short• Second beat long
Mazurka in A minorOp. 17, No. 4
measure with longer second beat
measure with longer first beat
• blurred image to show overall structure
B AA A D(A) C
![Page 20: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/20.jpg)
Standard Deviation & Variance
Tempo
Duration
Variance:
StandardDeviation: Tempo
Duration
phrasing
phrasing
mazurka script
mazurka script
longer durations = more variability
some performers do it, others don’t
![Page 21: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/21.jpg)
Timescapes
• Examine the internal tempo structure of a performances
• Plot average tempos over various time-spans in the piece
• Example of a piece with 6 beats at tempos A, B, C, D, E, and F:
average tempo forentire piece
plot of individualtempos
average tempo of adjacent neighbors
3-neighbor average
4-neighbor average
5-neighbor average
![Page 22: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/22.jpg)
Timescapes (2)average tempo of performance
averagefor performance
slower
faster
phrases
![Page 23: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/23.jpg)
Comparison of performers
6
![Page 24: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/24.jpg)
Same performer
![Page 25: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/25.jpg)
Correlation
Pearson correlation:
• Measures how well two shapes match:
r = 1.0 is an exact match.r = 0.0 means no relation at all.
![Page 26: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/26.jpg)
Overall performance correlations
BiretBrailowsky
ChiuFriereIndjic
LuisadaRubinstein 1938Rubinstein 1966
SmithUninsky
Bi LuBr Ch Fl In R8 R6 Sm Un
Highest correlationto Biret 1990
Lowest correlationto Biret 1990
![Page 27: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/27.jpg)
Correlation tree• Who is closest to whom?
(with respect to beat temposof an entire performance).
Mazurka in A minor, 68/3
![Page 28: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/28.jpg)
Correlation tree (2)Mazurka in A minor, 17/4
![Page 29: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/29.jpg)
Correlation network• How close is everyone to everyone else?
Mazurka in A minor, 17/4
![Page 30: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/30.jpg)
Correlation scapes• Who is most similar to a particular performer at any given region in the music?
![Page 31: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/31.jpg)
Same performer over time3 performances by Rubinstein of mazurka 17/4 in A minor
(30 performances compared)
![Page 32: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/32.jpg)
Same performer (2)2 performances by Horowitz of mazurka 17/4 in A minor plus Biret 1990 performance.
(30 performances compared)
![Page 33: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/33.jpg)
Student/Teacher
• Francois and Biret both studied with Cortot,
Mazurka in F major 68/3
(20 performances compared)
![Page 34: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/34.jpg)
Correlation to average
![Page 35: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/35.jpg)
Possible influences
![Page 36: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/36.jpg)
Same source recordingThe same performance by Magaloff on two different CD releases
Philips 456 898-2 Philips 426 817/29-2
• Structures at bottoms due to errors in beat extraction or interpreted beat locations (no notes on the beat).
mazurka 17/4 in A minor
![Page 37: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/37.jpg)
Purely coincidentalTwo difference performances from two different performers on two different record labels from two different countries.
![Page 38: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/38.jpg)
Arch Correlation
Paderewski 1912 (17-4)
red = phrase peakblue = phrase boundary
make your own plots at http://mazurka.org.uk/software/online/scape
![Page 39: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/39.jpg)
Phrase Measure Level
Average performance (17-4)
A AB
A C A
DC1 C2 C3
structural boundaries
![Page 40: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/40.jpg)
Phrase Identification/Characterization
Phrase Edges
Ashkenazy c.1980 (17-4)
Chiu 1999
Rosen 1989
Paderewski 1912
![Page 41: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/41.jpg)
Arch Correlation (2)Rubinstein 1938
Rubinstein 1952
Rubinstein 1966
Horowitz 1971
Horowitz 1985
![Page 42: Mazurka Project Update](https://reader035.vdocuments.net/reader035/viewer/2022081420/56814366550346895dafe14c/html5/thumbnails/42.jpg)
Ramp Correlation
average 17-4
Arch
Ramp
red = accel.blue = rit.