A Quantum Kernel Learning Approach to Acoustic Modeling for Spoken Command Recognition