Gestural input effects on the spectral envelope of violin sounds - - PowerPoint PPT Presentation

gestural input effects on the spectral envelope of violin
SMART_READER_LITE
LIVE PREVIEW

Gestural input effects on the spectral envelope of violin sounds - - PowerPoint PPT Presentation

Gestural input effects on the spectral envelope of violin sounds Quim Llimona MUMT 605 Final Project December 18th, 2015 Background: MUSMAP Background: MUSMAP Background: MUSMAP What does a violin note look like? 0.1 0.02 Waveform


slide-1
SLIDE 1

Gestural input effects


  • n the spectral envelope 

  • f violin sounds

Quim Llimona

MUMT 605 Final Project

December 18th, 2015

slide-2
SLIDE 2

Background: MUSMAP

slide-3
SLIDE 3

Background: MUSMAP

slide-4
SLIDE 4

Background: MUSMAP

slide-5
SLIDE 5

What does a violin note look like?

0.2 0.4 0.6 0.8 1 −0.1 0.1

Waveform Spectrogram

0.2 0.4 0.6 0.8 1 5 10 15 20 0.2 0.4 0.6 0.8 1 0.2 0.4

Bow velocity

0.2 0.4 0.6 0.8 1 0.35 0.4 0.45 0.5

Bow force

0.2 0.4 0.6 0.8 1 −0.04 −0.02 0.02

Waveform Spectrogram

0.2 0.4 0.6 0.8 1 5 10 15 20 0.2 0.4 0.6 0.8 1 0.5 1

Bow velocity

0.2 0.4 0.6 0.8 1 0.5 1

Bow force

slide-6
SLIDE 6

What does a violin note look like?

0.2 0.4 0.6 0.8 1 −0.1 0.1

Waveform Spectrogram

0.2 0.4 0.6 0.8 1 5 10 15 20 0.2 0.4 0.6 0.8 1 0.2 0.4

Bow velocity

0.2 0.4 0.6 0.8 1 0.35 0.4 0.45 0.5

Bow force

0.2 0.4 0.6 0.8 1 −0.04 −0.02 0.02

Waveform Spectrogram

0.2 0.4 0.6 0.8 1 5 10 15 20 0.2 0.4 0.6 0.8 1 0.5 1

Bow velocity

0.2 0.4 0.6 0.8 1 0.5 1

Bow force

slide-7
SLIDE 7

Spectral envelope extraction: LPC

0.5 1 1.5 2 x 10

4

−80 −60 −40 −20 20 40

LPC (p=8) Frequency (Hz) Power (dB)

0.5 1 1.5 2 x 10

4

−80 −60 −40 −20 20 40

LPC (p=16) Frequency (Hz) Power (dB)

0.5 1 1.5 2 x 10

4

−80 −60 −40 −20 20 40

LPC (p=32) Frequency (Hz) Power (dB)

0.5 1 1.5 2 x 10

4

−80 −60 −40 −20 20 40

LPC (p=64) Frequency (Hz) Power (dB)

slide-8
SLIDE 8

Spectral envelope extraction: PEAKPICK

0.5 1 1.5 2 x 10

4

−80 −60 −40 −20 20 40

Frequency (Hz) Magnitude (dB)

slide-9
SLIDE 9

Spectral envelope extraction: MAXPOOL

0.5 1 1.5 2 x 10

4

−80 −60 −40 −20 20 40

MAXPOOL (n=218) Frequency (Hz) Power (dB)

0.5 1 1.5 2 x 10

4

−80 −60 −40 −20 20 40

MAXPOOL (n=72) Frequency (Hz) Power (dB)

0.5 1 1.5 2 x 10

4

−80 −60 −40 −20 20 40

MAXPOOL (n=55) Frequency (Hz) Power (dB)

0.5 1 1.5 2 x 10

4

−80 −60 −40 −20 20 40

MAXPOOL (n=22) Frequency (Hz) Power (dB)

slide-10
SLIDE 10

Spectral whitening


Admittance deconvolution

0.5 1 1.5 2 x 10

4

−40 −20 20 40 60 80

Deconvolved

Frequency (Hz) Power (dB)

0.5 1 1.5 2 x 10

4

−80 −60 −40 −20 20 40

Raw

Frequency (Hz) Power (dB)

slide-11
SLIDE 11

Spectral whitening

Data-driven normalization

10

3

10

4

−60 −40 −20 20

Frequency (Hz) Magnitude (dB)

10

3

10

4

−60 −40 −20 20

Frequency (Hz) Magnitude (dB)

10

3

10

4

−60 −40 −20 20

Frequency (Hz) Magnitude (dB)

10

3

10

4

−60 −40 −20 20

Frequency (Hz) Magnitude (dB)

Original Whitened Original Whitened Original Whitened Original Whitened

slide-12
SLIDE 12

Original Frequency (Hz) 0.0 200 522 1361 3552 9267 0.2 0.4 0.6 0.8 1 1.2 5 10 15 Mean bow force Frequency (Hz) Deconvolved 0.0 200 522 1361 3552 9267 Original Frequency (Hz) 0.2 200 522 1361 3552 9267 0.4 0.6 0.8 1 1.2 1.4 5 10 15 20 Mean bow velocity Frequency (Hz) Deconvolved 0.2 200 522 1361 3552 9267 Original Frequency (Hz) 9 200 522 1361 3552 9267 20 30 40 50 5 10 15 Mean bow bridge distance Frequency (Hz) Deconvolved 9 200 522 1361 3552 9267

slide-13
SLIDE 13

10

3

10

4

0.2 0.4 0.6 0.8 1

Frequency (Hz)

Learn convolutive templates through Non-Negative Matrix Factorization (NMF)

  • n the log-spectrum (where chained filters are additive)
slide-14
SLIDE 14

Distribution of NMF activations

slide-15
SLIDE 15

Distribution of NMF activations

slide-16
SLIDE 16

Distribution of NMF activations

slide-17
SLIDE 17

Ratio and sum of NMF activations

slide-18
SLIDE 18

Thanks!

Quim Llimona

MUMT 605 Final Project

December 18th, 2015