notesum.ai

Published at October 22

Can a Machine Distinguish High and Low Amount of Social Creak in Speech?

cs.RO

cs.AI

Released Date: October 22, 2024

Authors: Anne-Maria Laukkanen¹, Sudarsana Reddy Kadiri², Shrikanth Narayanan², Paavo Alku³

Aff.: ¹Speech and Voice Research Laboratory, Tampere University, Finland; ²Signal Analysis and Interpretation Laboratory (SAIL), University of Southern California, USA; ³Department of Information and Communications Engineering, Aalto University, Finland

Arxiv: https://arxiv.org/abs/2410.17028v1

Classifier	Hyperparameters
SVM (linear)	C=1.0
SVM (RBF)	C=1.0, gamma=0.1
Random Forest (RF)	n_estimators=100, max_depth=None, random_state=0
Multilayer Perceptron (MLP)	solver=adam, alpha=0.01, hidden_layer_sizes=(100,)
Logistic Regression (LR)	solver=lbfgs, C=1.0
Decision Tree (DT)	max_depth=5
AdaBoost	n_estimators=100, learning_rate=1.0