Skip to main content
Fig. 5 | EURASIP Journal on Audio, Speech, and Music Processing

Fig. 5

From: Efficient music identification using ORB descriptors of the spectrogram image

Fig. 5

Tempo test for robustness (Shazam vs ORB). This figure shows the results for the tempo being changed by a range from −50 to +50% to test the limits. This type of distortion is known as time stretching or time scaling. The speed of the audio clip is altered while the pitch remains constant. On the spectrogram, this change would be seen on the time axis (temporally) but no change on the frequency axis (pitch). The experiment was carried out where a random 20-s sample was taken from each song in the database, and the tempo change was applied. The distorted samples were then queried to the test if they accurately matched. It was seen that the Shazam algorithm produced better results compared to the ORB prototype

Back to article page