Figure 2.
(A) Raw audio data sample of one singer: no load condition 1 (top), low load condition (middle), high load condition (bottom). (B) Audio track samples after the noise below −40 dB was removed from the three signals of Figure 2A. (C) An example of raw audio (the first layer corresponds with first layer of Figure 2A) and after noise removal (the second layer corresponds with the first layer of Figure 2B), with the silent regions extracted (the third layer is based on the data of the second layer). Finally, the fourth layer shows the silent regions that are taken into account after a manual inspection of the analysis.