**5. Conclusions**

In this chapter, a general review of the common classification and separation algorithms used for speech and music was presented and some were introduced and discussed thoroughly. The approaches dealt with classification were divided into three categories. The first category included most of the real-time approaches. In the real-time approaches, we introduced the ZCR, the STE, the ZCR and the STE with positive derivative, with some of their modified versions, and the neural networks. The second category included most of the frequency domain approaches such as the spectral centroid and its variance, the spectral flux and its variance, the roll-off of the spectrum, the cepstral residual, and the delta pitch. However, the last category introduced two time-frequency approaches, mainly the spectrogram and the evolutionary spectrum. It has been noticed that the time-frequency classifiers provided an excellent and a robust discrimination result in discriminating speech from music signals in digital audio. Depending on the application, the decision of which feature should be chosen is selected. The algorithms of the first category are faster since the processing is made in the real time; however, those of the second


#### **Table 9.**

*Summary of the classification and separation algorithms.*

*Classification and Separation of Audio and Music Signals DOI: http://dx.doi.org/10.5772/intechopen.94940*

one are more precise. The time-frequency approaches has not been discussed thoroughly in literature and they still need more research and elaboration. Lastly, we may conclude that many classification algorithms were proposed in literature, however, few ones were proposed for separation. The algorithms introduced in this chapter can be summarized in **Table 9**.
