Music Classification by Genre: Overall Results
Summary: When tested with the training vectors, the system is 87.5% accurate. Higher accuracy implies that the system has memorized the training set and is unable to generalize when given new inputs.
Bandwidth Figure 1:
Overall, bandwidth is a good detector for jazz and rap, but poorer in distinguishing between classical, punk, techno, and country, which all have about the same bandwidth.
|
Beat (Tempo) Figure 2:
Rap is the only genre that can be distinguished by its beat. Classical and jazz have much higher variability in tempo, since they often consist of a long piece subdivided into sections.
|
Frequency Cutoff Figure 3:
Though difficult to isolate any one genre, this analysis tool does separate them into two main categories: (1) Classical, Punk and Country (2) Techno, Jazz, and Rap.
|
Frequency Smoothness Figure 4:
While freqsmooth does give a different value for each genre, it also gives a radically different value for songs within a given genre. In other words, it does not give a good representation of a genre as a whole. Given the plus and minus standard deviation bars, each genre overlaps heavily.
|
High Pass Filtering Figure 5:
Like one would expect, classical had the smallest error of any genre tested. This should be intuitive since it uses the lower frequency part of the spectrum. One can think of classical music as being very fluid with few sudden changes in frequency. Conversely, punk and jazz had the highest amount of error, which is a good indication of higher frequencies being utilized. Compared to classical music, these genres are much less fluid and often exhibit rapid changes in tempo. Somewhere between these two extremes are techno, rap and country.
|
Total Power Figure 6:
The standard deviations of rap and techno are very distinct, whereas the others are all about the same value. Although the average total power of techno may not be a good indicator, the standard deviation should be able to pick out techno.
|
Comments, questions, feedback, criticisms?
Discussion forum
Send feedback
More about this content: Metadata | Version History | Cite This Content
This work is licensed by Melodie Chu, Christopher Hunter, and Mitali Banerjee. See the Creative Commons License about permission to reuse this material.
Last edited by Mitali Banerjee on Dec 11, 2003 9:36 pm US/Central.