Improved Frame-Wise Segmentation of Audio Signals for Smart Hearing Aid Using Particle Swarm Optimization-Based Clustering

Algorithm 1

PSO-based clustering for audio speech signals.

	Initialize signals with a constant value of energy and deploy in a specified area.
	Initially, assign 10% of n signals to cluster nodes at random.
	For i: 1 to n
	Calculate distance (F1) = di, m + 1
	Minimum distance = Euclidean distance value from signal i to BSm + 1
	If (minimum distance > distance)
	Node.id = i
	End if
	For j: 1 to m (total number of CH)
	Calculate distance (F2) = djm (distance from signal j to CHm)
	If (minimum distance > distance)
	Minimum distance = distance
	Node.id = j
	End if
	Store the distance in an array (A) which maintains values for clusters
	A (Node.id).sum = A (Node.id).sum + minimum_distance
	A (Node.id).num = A.(Node.id).num + 1
	End for
	End for
	For k: 1 to m (total number of CH)
	Calculate distance (F3) = dk, m + 1
	If (minimum distance > distance)
	Minimum distance = distance
	Cluster.id = k
	End if
	Store the distance in an array (A) which maintains values for clusters
	A (cluster.id).sum = A (cluster.id).sum + minimum_distance
	A (cluster.id).num = A(cluster.id).num + 1
	End for
	Calculate the total energy
	Fitness function value for each node
	For i: 1 to n
	Fitness (clustering) = (0.25 ∗ F1) + (0.25 ∗ F2) + (0.25 ∗ F3) + (0.25 ∗ F4)
	End for
	Stop