Improved Child Text-to-Speech Synthesis through Fastpitch-based Transfer Learning