Tabla Strokes Recognition

1
Tabla Strokes Recognition

• Mihir Sarkar

2
Tabla?
3
Context

• Can you distinguish different bols?
• Can a machine automatically classify tabla
  strokes?
• Is there a systematic way to identify the best
  method to recognize tabla strokes?

4
Vision
5
Experimental Setup

• 1 tabla set
• 3 tabla players
• 10 bols
• 413 recordings (kept 300)
• Microphone input (studio recording)
• Discrete strokes

6
Raw data
7
Spectrogram
8
Feature Extraction

• Time domain ZCR
• Frequency domain PSD
• Cepstral domain MFCC

9
Dataset Selection

• Orthogonal dimensions
• Instances
• Bols
• Players
• Training / leave-one-out validation
• Testing

10
Baseline

• Random 10
• Human 87
• Initial k-NN 18(Welchs PSD, NFFT 16, k 1)

11
k-NN
12
k-NN
13
k-NN
14
k-NN
15
k-NN
16
k-NN
17
Confusion Matrix
Na Tin Ga Ka Dha Dhin Te Re Tat Thun
Na 5 0 0 0 0 0 0 0 0 1
Tin 0 3 0 0 1 0 0 0 0 2
Ga 0 0 4 0 0 2 0 0 0 0
Ka 0 1 0 3 0 1 0 1 0 0
Dha 0 0 2 0 2 1 1 0 0 0
Dhin 0 0 1 0 1 4 0 0 0 0
Te 0 1 0 0 0 0 1 0 4 0
Re 0 0 0 0 0 0 1 4 0 1
Tat 0 0 0 0 0 0 2 1 3 0
Thun 0 1 0 0 0 0 0 0 0 5
18
Neural Networks
Nodes 20 22 24 36 40 42 46 48 50 52
Validation 53 47 13 41 55 70 40 40 45 72
Testing 5 13 15 10 18 12 6.7 10 10 10
19
Contributions

• Implemented pattern classification algorithms
  (Matlab)
• Analyzed recognition rates with varying
  parameters
• Explored a systematic way to perform
  classification

20
Future Directions

• Vibration sensors
• More recordings
• Timing (multiple frames, HMM)
• Real-time
• Continuous strokes
• Integrate context (rhythmic patterns)