Open Access

Research on automatic identification and evaluation method of piano playing skills based on convolutional neural network

Xiaoliang Wu

Xiaoliang Wu

School of Contemporary Music, Shandong University of ArtsJinan, China
Search for this author on
Sciendo|Google Scholar
Wu, Xiaoliang
  
Mar 21, 2025
Research on automatic identification and evaluation method of piano playing skills based on convolutional neural network's Cover Image
About this article
Previous Article
Next Article
Cite
Download Cover
Published Online: Mar 21, 2025
Received: Oct 20, 2024
Accepted: Feb 02, 2025
DOI: https://doi.org/10.2478/amns-2025-0622
Keywords
© 2025 Xiaoliang Wu, published by Sciendo
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Figure 1.

System framework
System framework

Figure 2.

System module
System module

Figure 3.

Scoring algorithm block diagram
Scoring algorithm block diagram

Figure 4.

The two Numbers are based on the collection
The two Numbers are based on the collection

Figure 5.

The curve of the piano playing hand changes
The curve of the piano playing hand changes

Figure 6.

The distribution of finger continuous keystroke time
The distribution of finger continuous keystroke time

Figure 7.

The result of the time domain waveform of the semi-tone order 4(fa)
The result of the time domain waveform of the semi-tone order 4(fa)

Figure 8.

Semi-tone order 4(fa) frequency domain results
Semi-tone order 4(fa) frequency domain results

Example of the left finger span

Finger number L5 L4 L3 L2 L1
L5 — 5.55 7.05 8.05 11.55
L4 5.55 — 4.05 7.05 10.55
L3 7.05 4.05 — 5.55 9.05
L2 8.05 7.05 5.55 — 8.55
L1 11.55 10.55 9.05 8.55 —

Example of the right finger span

Finger number R1 R2 R3 R4 R5
R1 -- 8.55 9.55 10.55 11.55
R2 8.55 -- 5.55 7.05 9.05
R3 9.55 5.55 -- 4.05 5.55
R4 10.55 7.05 4.05 -- 4.55
R5 11.55 10.55 9.05 8.55 —

The finger is a continuous stroke and a flexible score

Finger number Mean Standard deviation Flexible average score
L5 1522.198 172.484 61.21
L4 1638.956 166.778 53.77
L3 1509.271 155.468 66.4
L2 1476.312 177.393 69.31
L1 1518.939 174.481 59.66
R1 1404.546 178.336 73
R2 1314.255 139.574 78.32
R3 1360.534 132.529 71.62
R4 1484.587 156.215 71.8
R5 1422.542 160.372 68.85

Analysis of the ability of different finger movements

Feature extraction The accuracy of the finger motion extraction(%)
Thumb Index Middle Ring Finger
Handback Angle deviation 97.095 96.575 93.418 99.424 97.213
Elevation aberration 98.43 98.287 98.075 96.356 96.326
Angle deviation 97.733 99.254 99.382 97.693 98.146
Cross Angle anomaly 95.92 100 99.026 100 95.432
Acceleration standard deviation 98.275 99.897 97.083 97.894 98.143
Standard deviation 96.713 99.615 99.171 95.663 99.642
Angular velocity 95.36 98.604 98.144 100 98.188
Lower finger Angle deviation 96.081 100 98.179 99.902 96.333
Elevation aberration 98.596 100 100 100 96.736
Angle deviation 95.946 98.202 96.977 97.918 97.544
Cross Angle anomaly 93.586 97.657 100 97.024 99.371
Acceleration standard deviation 95.741 97.362 98.857 97.367 97.616
Standard deviation 98.6 100 100 97.714 96.094
Angular velocity 94.91 95.79 99.519 98.123 97.329
Knuckle Angle deviation 97.847 98.066 99.5 96.653 96.642
Elevation aberration 96.563 98.21 97.891 97.161 100
Angle deviation 96.883 98.891 97.3 96.603 98.627
Cross Angle anomaly 98.727 99.909 97.421 99.085 98.224
Acceleration standard deviation 97.777 98.935 96.007 96.22 97.338
Standard deviation 98.94 98.383 96.957 96.978 95.886
Angular velocity 98.92 98.95 98.553 97.403 100

Video assessment experiment results

Subset Tick Holder Erase Big pinch Little pinch
Accuracy rate(%) 95.55 96.77 97.53 90.49 85.42

Audio comparison experimental results

Subset Tick Holder Erase Big pinch Little pinch
Accuracy rate(%) 99.02 99.97 98.62 96.03 97.46
Language:
English
Publication timeframe:
1 times per year
Journal Subjects:
Life Sciences, Life Sciences, other, Mathematics, Applied Mathematics, General Mathematics, Physics, Physics, other
Journal RSS Feed